diff -urN linux-2.4.7/Documentation/Configure.help linux-2.4.7-mcore/Documentation/Configure.help --- linux-2.4.7/Documentation/Configure.help Thu Jul 19 20:48:15 2001 +++ linux-2.4.7-mcore/Documentation/Configure.help Fri Jul 27 14:58:14 2001 @@ -17569,6 +17569,35 @@ boards supported by this driver, and for further information on the use of this driver. +Boot kernel image support +CONFIG_BOOTIMG + Add support for booting a new Linux kernel from a running Linux + system. You need to download the bootimg(8) utility from + ftp://icaftp.epfl.ch/pub/people/almesber/misc/bootimg-current.tar.gz + in order to use this functionality. + +Protect SMP configuration tables +CONFIG_BOOTIMG_SMP + On SMP systems, the BIOS stores tables with configuration data in + memory and an SMP-enabled kernel reads these tables. However, a + kernel without SMP support will overwrite such tables. If a kernel + without SMP support used bootimg to boot an SMP-enabled kernel, the + latter will probably crash when trying to read the SMP tables. The + CONFIG_BOOTIMG_SMP option enables minimal support for scanning and + protecting of SMP configuration tables also for kernels without SMP + support. + +In-memory kernel core dump facility +CONFIG_MCL_COREDUMP + In conjunction with bootimg, this allows you to get kernel core dumps + of your system at panic() time. The panic call is modified so that it + calls the core dump facility and reboots the system. On the way back + up, the kernel dump image is written out to disk by the accompanying + init script. You can use the crash analysis tool to analyze the core + dump. This tool can be found at : + + http://www.missioncriticallinux.com/download + # # ARM options # diff -urN linux-2.4.7/Makefile linux-2.4.7-mcore/Makefile --- linux-2.4.7/Makefile Fri Jul 20 01:03:56 2001 +++ linux-2.4.7-mcore/Makefile Fri Jul 27 15:00:19 2001 @@ -16,7 +16,7 @@ FINDHPATH = $(HPATH)/asm $(HPATH)/linux $(HPATH)/scsi $(HPATH)/net HOSTCC = gcc -HOSTCFLAGS = -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer +HOSTCFLAGS = -Wall -Wstrict-prototypes -O2 -g CROSS_COMPILE = @@ -88,8 +88,14 @@ CPPFLAGS := -D__KERNEL__ -I$(HPATH) CFLAGS := $(CPPFLAGS) -Wall -Wstrict-prototypes -Wno-trigraphs -O2 \ - -fomit-frame-pointer -fno-strict-aliasing -fno-common + -fno-strict-aliasing -fno-common AFLAGS := -D__ASSEMBLY__ $(CPPFLAGS) + +ifeq ($(CONFIG_MCL_COREDUMP),y) + CFLAGS += -g +else + CFLAGS += -fomit-frame-pointer +endif # # ROOT_DEV specifies the default root-device when making the image. diff -urN linux-2.4.7/arch/alpha/config.in linux-2.4.7-mcore/arch/alpha/config.in --- linux-2.4.7/arch/alpha/config.in Tue Jul 10 23:08:51 2001 +++ linux-2.4.7-mcore/arch/alpha/config.in Fri Jul 27 14:58:14 2001 @@ -364,5 +364,6 @@ bool 'Magic SysRq key' CONFIG_MAGIC_SYSRQ bool 'Legacy kernel start address' CONFIG_ALPHA_LEGACY_START_ADDRESS +bool 'Kernel Core Dump Facility' CONFIG_MCL_COREDUMP endmenu diff -urN linux-2.4.7/arch/alpha/kernel/Makefile linux-2.4.7-mcore/arch/alpha/kernel/Makefile --- linux-2.4.7/arch/alpha/kernel/Makefile Fri Feb 9 14:38:01 2001 +++ linux-2.4.7-mcore/arch/alpha/kernel/Makefile Fri Jul 27 14:58:14 2001 @@ -45,6 +45,9 @@ else +# MCLX crash dump support +obj-$(CONFIG_MCL_COREDUMP) += crash.o + # Core logic support obj-$(CONFIG_ALPHA_APECS) += core_apecs.o obj-$(CONFIG_ALPHA_CIA) += core_cia.o diff -urN linux-2.4.7/arch/alpha/kernel/crash.c linux-2.4.7-mcore/arch/alpha/kernel/crash.c --- linux-2.4.7/arch/alpha/kernel/crash.c Wed Dec 31 19:00:00 1969 +++ linux-2.4.7-mcore/arch/alpha/kernel/crash.c Fri Jul 27 14:58:14 2001 @@ -0,0 +1,130 @@ +/* + * linux/arch/alpha/crash.c + * + * Architecture dependant code for MCL in-memory core dump. + */ +#include +#include +#include +#include +#include +#include + +unsigned long crash_console_sp = 0; + +inline void crash_save_regs(void) { + static unsigned long gprs[32]; + + __asm__ __volatile__("stq $0,%0" : "=m"(gprs[0])); + __asm__ __volatile__("stq $1,%0" : "=m"(gprs[1])); + __asm__ __volatile__("stq $2,%0" : "=m"(gprs[2])); + __asm__ __volatile__("stq $3,%0" : "=m"(gprs[3])); + __asm__ __volatile__("stq $4,%0" : "=m"(gprs[4])); + __asm__ __volatile__("stq $5,%0" : "=m"(gprs[5])); + __asm__ __volatile__("stq $6,%0" : "=m"(gprs[6])); + __asm__ __volatile__("stq $7,%0" : "=m"(gprs[7])); + __asm__ __volatile__("stq $8,%0" : "=m"(gprs[8])); + __asm__ __volatile__("stq $9,%0" : "=m"(gprs[9])); + __asm__ __volatile__("stq $10,%0" : "=m"(gprs[10])); + __asm__ __volatile__("stq $11,%0" : "=m"(gprs[11])); + __asm__ __volatile__("stq $12,%0" : "=m"(gprs[12])); + __asm__ __volatile__("stq $13,%0" : "=m"(gprs[13])); + __asm__ __volatile__("stq $14,%0" : "=m"(gprs[14])); + __asm__ __volatile__("stq $15,%0" : "=m"(gprs[15])); + __asm__ __volatile__("stq $16,%0" : "=m"(gprs[16])); + __asm__ __volatile__("stq $17,%0" : "=m"(gprs[17])); + __asm__ __volatile__("stq $18,%0" : "=m"(gprs[18])); + __asm__ __volatile__("stq $19,%0" : "=m"(gprs[19])); + __asm__ __volatile__("stq $20,%0" : "=m"(gprs[20])); + __asm__ __volatile__("stq $21,%0" : "=m"(gprs[21])); + __asm__ __volatile__("stq $22,%0" : "=m"(gprs[22])); + __asm__ __volatile__("stq $23,%0" : "=m"(gprs[23])); + __asm__ __volatile__("stq $24,%0" : "=m"(gprs[24])); + __asm__ __volatile__("stq $25,%0" : "=m"(gprs[25])); + __asm__ __volatile__("stq $26,%0" : "=m"(gprs[26])); + __asm__ __volatile__("stq $27,%0" : "=m"(gprs[27])); + __asm__ __volatile__("stq $28,%0" : "=m"(gprs[28])); + __asm__ __volatile__("stq $29,%0" : "=m"(gprs[29])); + __asm__ __volatile__("stq $30,%0" : "=m"(gprs[30])); + __asm__ __volatile__("stq $31,%0" : "=m"(gprs[31])); + + panic_regs = gprs; +} + +/* + * Save the current stack pointer. + */ +void crash_save_current_state(struct task_struct *tp) +{ + register unsigned long sp __asm__("$30"); + tp->thread.ksp = sp; + panic_ksp[smp_processor_id()] = sp; + mb(); + + save_core(); + + /* save_core only returns if we're running on the boot cpu */ + crash_halt_or_reboot(1); +} + +/* + * If we are not the panicking thread, we simply halt. Otherwise, + * we take care of calling the reboot code. + */ +void crash_halt_or_reboot(int boot_cpu) +{ + unsigned long *pflags, flags; + struct percpu_struct *cpup; + int cpuid = smp_processor_id(); + + cpup = (struct percpu_struct *) + ((unsigned long)hwrpb + hwrpb->processor_offset + + hwrpb->processor_size * cpuid); + + pflags = &cpup->flags; + flags = *pflags; + +#ifdef CONFIG_SMP + if (!boot_cpu) { + flags |= 0x00040000UL; /* "halt machine" */ + *pflags = flags; + clear_bit(cpuid, &cpu_present_mask); + halt(); + } + else { +#else + { +#endif + printk("crash_halt_or_reboot: "); + if (console_crash) { + printk("halting system. Type 'b' to boot.\n"); + halt(); + } + printk("restarting system.\n"); + machine_restart(NULL); + } +} + +void crash_cleanup_smp_state(void) +{ + /* + * Not much to be done here for alpha. + */ + __cli(); +} + +/* + * Crash IPI + */ +void smp_crash_funnel_cpu(void) +{ + crash_save_current_state(current); +} + +void crash_callin(void) +{ + console_crash = 1; + smp_call_function((void*)smp_crash_funnel_cpu,0,0,0); + crash_save_current_state(current); + /* NOTREACHED */ +} diff -urN linux-2.4.7/arch/alpha/kernel/head.S linux-2.4.7-mcore/arch/alpha/kernel/head.S --- linux-2.4.7/arch/alpha/kernel/head.S Mon Apr 24 16:39:34 2000 +++ linux-2.4.7-mcore/arch/alpha/kernel/head.S Fri Jul 27 14:58:14 2001 @@ -55,6 +55,21 @@ .end __smp_callin #endif /* CONFIG_SMP */ +#ifdef CONFIG_MCL_COREDUMP + .align 3 + .globl __crash_callin + .ent __crash_callin +__crash_callin: + .prologue 1 + ldgp $29,0($27) + lda $9,crash_console_sp + ldq $10,0($9) + lda $30, 0x2000($10) /* one page of stack */ + jsr $26,crash_callin + call_pal PAL_halt + .end __crash_callin + +#endif /* CONFIG_MCL_COREDUMP */ # # The following two functions are needed for supporting SRM PALcode # on the PC164 (at least), since that PALcode manages the interrupt diff -urN linux-2.4.7/arch/alpha/kernel/process.c linux-2.4.7-mcore/arch/alpha/kernel/process.c --- linux-2.4.7/arch/alpha/kernel/process.c Wed Jun 20 14:10:27 2001 +++ linux-2.4.7-mcore/arch/alpha/kernel/process.c Fri Jul 27 14:58:14 2001 @@ -119,16 +119,20 @@ flags &= ~0x00ff0001UL; #ifdef CONFIG_SMP - /* Secondaries halt here. */ - if (cpuid != boot_cpuid) { - flags |= 0x00040000UL; /* "remain halted" */ - *pflags = flags; - clear_bit(cpuid, &cpu_present_mask); - halt(); - } + if (how->mode != LINUX_REBOOT_CMD_COREDUMP) { + /* Secondaries halt here. */ + if (cpuid != boot_cpuid) { + flags |= 0x00040000UL; /* "remain halted" */ + *pflags = flags; + clear_bit(cpuid, &cpu_present_mask); + halt(); + } + } #endif - if (how->mode == LINUX_REBOOT_CMD_RESTART) { + if (how->mode == LINUX_REBOOT_CMD_RESTART || + how->mode == LINUX_REBOOT_CMD_COREDUMP) { + if (!how->restart_cmd) { flags |= 0x00020000UL; /* "cold bootstrap" */ } else { @@ -148,10 +152,12 @@ *pflags = flags; #ifdef CONFIG_SMP - /* Wait for the secondaries to halt. */ - clear_bit(boot_cpuid, &cpu_present_mask); - while (cpu_present_mask) - barrier(); + if (how->mode != LINUX_REBOOT_CMD_COREDUMP) { + /* Wait for the secondaries to halt. */ + clear_bit(boot_cpuid, &cpu_present_mask); + while (cpu_present_mask) + barrier(); + } #endif /* If booted from SRM, reset some of the original environment. */ @@ -180,6 +186,7 @@ halt(); } +extern char *panicmsg; static void common_shutdown(int mode, char *restart_cmd) { @@ -187,7 +194,8 @@ args.mode = mode; args.restart_cmd = restart_cmd; #ifdef CONFIG_SMP - smp_call_function(common_shutdown_1, &args, 1, 0); + if(!panicmsg) + smp_call_function(common_shutdown_1, &args, 1, 0); #endif common_shutdown_1(&args); } @@ -195,6 +203,9 @@ void machine_restart(char *restart_cmd) { + if (panicmsg) + common_shutdown(LINUX_REBOOT_CMD_COREDUMP, restart_cmd); + else common_shutdown(LINUX_REBOOT_CMD_RESTART, restart_cmd); } diff -urN linux-2.4.7/arch/alpha/kernel/setup.c linux-2.4.7-mcore/arch/alpha/kernel/setup.c --- linux-2.4.7/arch/alpha/kernel/setup.c Thu May 24 18:20:18 2001 +++ linux-2.4.7-mcore/arch/alpha/kernel/setup.c Fri Jul 27 14:58:14 2001 @@ -56,6 +56,9 @@ #include "proto.h" #include "pci_impl.h" +#ifdef CONFIG_MCL_COREDUMP +#include +#endif struct hwrpb_struct *hwrpb; unsigned long srm_hae; @@ -241,7 +244,7 @@ struct memdesc_struct * memdesc; unsigned long start_kernel_pfn, end_kernel_pfn; unsigned long bootmap_size, bootmap_pages, bootmap_start; - unsigned long start, end; + unsigned long start, end, crash_pages=0; int i; /* Find free clusters, and init and free the bootmem accordingly. */ @@ -286,6 +289,9 @@ bootmap_pages = bootmem_bootmap_pages(max_low_pfn); /* Now find a good region where to allocate the bootmap. */ +#ifdef CONFIG_MCL_COREDUMP + crash_pages = crash_pages_needed(); +#endif for_each_mem_cluster(memdesc, cluster, i) { if (cluster->usage & 3) continue; @@ -305,7 +311,7 @@ end = start_kernel_pfn; } else if (start < end_kernel_pfn) start = end_kernel_pfn; - if (end - start >= bootmap_pages) { + if (end - start >= (bootmap_pages + crash_pages)) { bootmap_start = start; break; } @@ -316,6 +322,11 @@ goto try_again; } +#ifdef CONFIG_MCL_COREDUMP + crash_init((u_long)phys_to_virt(PFN_PHYS(bootmap_start)), + (u_long)phys_to_virt(PFN_PHYS(bootmap_start + bootmap_pages)), + (u_long)phys_to_virt(PFN_PHYS(bootmap_start + bootmap_pages + crash_pages))); +#endif /* Allocate the bootmap and mark the whole MM as reserved. */ bootmap_size = init_bootmem(bootmap_start, max_low_pfn); @@ -349,8 +360,18 @@ printk("freeing pages %ld:%ld\n", start, end); } - /* Reserve the bootmap memory. */ + /* Reserve the bootmap and crash memory. */ +#ifndef CONFIG_MCL_COREDUMP reserve_bootmem(PFN_PHYS(bootmap_start), bootmap_size); +#else + reserve_bootmem(PFN_PHYS(bootmap_start), bootmap_size + + crash_pages * PAGE_SIZE); + + /* reserve crash pages */ + crash_mark_dump_reserved(); + crash_console_sp = (unsigned long)alloc_bootmem(PAGE_SIZE) | + 0xfffffc0000000000; +#endif printk("reserving pages %ld:%ld\n", bootmap_start, bootmap_start+PFN_UP(bootmap_size)); #ifdef CONFIG_BLK_DEV_INITRD diff -urN linux-2.4.7/arch/alpha/kernel/smp.c linux-2.4.7-mcore/arch/alpha/kernel/smp.c --- linux-2.4.7/arch/alpha/kernel/smp.c Mon Jun 11 22:15:27 2001 +++ linux-2.4.7-mcore/arch/alpha/kernel/smp.c Fri Jul 27 14:58:14 2001 @@ -35,6 +35,9 @@ #include "proto.h" #include "irq_impl.h" +#ifdef CONFIG_MCL_COREDUMP +#include +#endif #define DEBUG_SMP 0 #if DEBUG_SMP @@ -647,6 +650,15 @@ ((bogosum + 2500) / (5000/HZ)) % 100); smp_num_cpus = cpu_count; + +#ifdef CONFIG_MCL_COREDUMP + /* + * Here we set things up so we can get a crash from the SRM console. + */ + hwrpb->CPU_restart = __crash_callin; /*__smp_callin;*/ + hwrpb->CPU_restart_data = (unsigned long) __crash_callin;/*__smp_callin;*/ + hwrpb_update_checksum(hwrpb); +#endif } /* diff -urN linux-2.4.7/arch/alpha/kernel/traps.c linux-2.4.7-mcore/arch/alpha/kernel/traps.c --- linux-2.4.7/arch/alpha/kernel/traps.c Wed Jul 4 14:50:38 2001 +++ linux-2.4.7-mcore/arch/alpha/kernel/traps.c Fri Jul 27 14:58:14 2001 @@ -15,6 +15,10 @@ #include #include +#ifdef CONFIG_MCL_COREDUMP +#include +#endif + #include #include #include @@ -536,6 +540,10 @@ while (1); } current->thread.flags |= (1UL << 63); +#ifdef CONFIG_MCL_COREDUMP + if(panic_on_oops) + panic("die_if_kernel"); +#endif do_exit(SIGSEGV); } diff -urN linux-2.4.7/arch/i386/boot/compressed/head.S linux-2.4.7-mcore/arch/i386/boot/compressed/head.S --- linux-2.4.7/arch/i386/boot/compressed/head.S Wed Jul 5 15:03:12 2000 +++ linux-2.4.7-mcore/arch/i386/boot/compressed/head.S Fri Jul 27 14:58:14 2001 @@ -31,6 +31,24 @@ startup_32: cld cli + +/* + * GDT is invalid if we're booted by bootimg, so reload it now + */ + lgdt %cs:gdt_descr + ljmp $(__KERNEL_CS),$1f + +gdt_descr: + .word 31 + .long gdt_table + +gdt_table: /* stolen from arch/i386/kernel/head.S */ + .quad 0x0000000000000000 /* NULL descriptor */ + .quad 0x0000000000000000 /* not used */ + .quad 0x00cf9a000000ffff /* 0x10 kernel 4GB code at 0x00000000 */ + .quad 0x00cf92000000ffff /* 0x18 kernel 4GB data at 0x00000000 */ + +1: movl $(__KERNEL_DS),%eax movl %eax,%ds movl %eax,%es @@ -92,7 +110,6 @@ cld rep movsl - popl %esi # discard the address popl %ebx # real mode pointer popl %esi # low_buffer_start @@ -124,5 +141,6 @@ movsl movl %ebx,%esi # Restore setup pointer xorl %ebx,%ebx - ljmp $(__KERNEL_CS), $0x100000 + movl $0x100000,%eax + jmpl *%eax move_routine_end: diff -urN linux-2.4.7/arch/i386/boot/setup.S linux-2.4.7-mcore/arch/i386/boot/setup.S --- linux-2.4.7/arch/i386/boot/setup.S Tue May 1 19:04:56 2001 +++ linux-2.4.7-mcore/arch/i386/boot/setup.S Fri Jul 27 14:58:14 2001 @@ -97,16 +97,22 @@ # flags, unused bits must be zero (RFU) bit within loadflags loadflags: LOADED_HIGH = 1 # If set, the kernel is loaded high +RELOADS_GDT = 2 # if set, kernel reloads GDT, such that + # boot loader does not have to provide + # GDT in a "safe" memory location CAN_USE_HEAP = 0x80 # If set, the loader also has set # heap_end_ptr to tell how much # space behind setup.S can be used for # heap purposes. # Only the loader knows what is free -#ifndef __BIG_KERNEL__ - .byte 0 -#else - .byte LOADED_HIGH +_FLAGS = 0 +#ifdef __BIG_KERNEL__ + _FLAGS = _FLAGS | LOADED_HIGH #endif +#ifdef CONFIG_BOOTIMG + _FLAGS = _FLAGS | RELOADS_GDT +#endif + .byte _FLAGS setup_move_size: .word 0x8000 # size to move, when setup is not # loaded at 0x90000. We will move setup diff -urN linux-2.4.7/arch/i386/config.in linux-2.4.7-mcore/arch/i386/config.in --- linux-2.4.7/arch/i386/config.in Sun Jul 15 19:15:44 2001 +++ linux-2.4.7-mcore/arch/i386/config.in Fri Jul 27 14:58:14 2001 @@ -390,4 +390,10 @@ #bool 'Debug kmalloc/kfree' CONFIG_DEBUG_MALLOC bool 'Magic SysRq key' CONFIG_MAGIC_SYSRQ +if [ "$CONFIG_EXPERIMENTAL" = "y" ]; then + bool 'Kernel Core Dump Facility' CONFIG_MCL_COREDUMP + if [ "$CONFIG_MCL_COREDUMP" = "y" ]; then + bool ' Reboot using bootimg' CONFIG_BOOTIMG + fi +fi endmenu diff -urN linux-2.4.7/arch/i386/kernel/Makefile linux-2.4.7-mcore/arch/i386/kernel/Makefile --- linux-2.4.7/arch/i386/kernel/Makefile Fri Dec 29 17:35:47 2000 +++ linux-2.4.7-mcore/arch/i386/kernel/Makefile Fri Jul 27 14:58:14 2001 @@ -40,5 +40,6 @@ obj-$(CONFIG_X86_LOCAL_APIC) += apic.o obj-$(CONFIG_X86_IO_APIC) += io_apic.o mpparse.o obj-$(CONFIG_X86_VISWS_APIC) += visws_apic.o +obj-$(CONFIG_MCL_COREDUMP) += crash.o include $(TOPDIR)/Rules.make diff -urN linux-2.4.7/arch/i386/kernel/crash.c linux-2.4.7-mcore/arch/i386/kernel/crash.c --- linux-2.4.7/arch/i386/kernel/crash.c Wed Dec 31 19:00:00 1969 +++ linux-2.4.7-mcore/arch/i386/kernel/crash.c Fri Jul 27 14:58:14 2001 @@ -0,0 +1,82 @@ +/* + * linux/arch/i386/crash.c + * + * Architecture dependant code for MCL in-memory core dump. + */ +#include +#include +#include +#include +#include +#include + +inline void crash_save_regs(void) { + static unsigned long regs[8]; + + __asm__ __volatile__("movl %%ebx,%0" : "=m"(regs[0])); + __asm__ __volatile__("movl %%ecx,%0" : "=m"(regs[1])); + __asm__ __volatile__("movl %%edx,%0" : "=m"(regs[2])); + __asm__ __volatile__("movl %%esi,%0" : "=m"(regs[3])); + __asm__ __volatile__("movl %%edi,%0" : "=m"(regs[4])); + __asm__ __volatile__("movl %%ebp,%0" : "=m"(regs[5])); + __asm__ __volatile__("movl %%eax,%0" : "=m"(regs[6])); + __asm__ __volatile__("movl %%esp,%0" : "=m"(regs[7])); + + panic_regs = regs; +} + +/* + * Save the current stack pointer and EIP. + */ +void crash_save_current_state(struct task_struct *tp) +{ + /* + * Here we save ebp instead of esp just in case the compiler + * decides to put an extra push in before we execute this + * instruction (thus invalidating our frame pointer). + */ + asm volatile("movl %%ebp,%0":"=m" (*(u_long *)&tp->thread.esp)); + tp->thread.eip = (u_long)crash_save_current_state; + panic_ksp[smp_processor_id()] = tp->thread.esp; + mb(); + + save_core(); + + crash_halt_or_reboot(1); +} + +/* + * If we are not the panicking thread, we simply halt. Otherwise, + * we take care of calling the reboot code. + */ +void crash_halt_or_reboot(int boot_cpu) +{ +#ifdef CONFIG_SMP + if (!boot_cpu) { + stop_this_cpu(NULL); + /* NOTREACHED */ + } +#endif + machine_restart(NULL); +} + +void crash_cleanup_smp_state(void) +{ + /* + * Here we duplicate smp_send_stop. Crash_halt_or_reboot() calls + * stop_this_cpu. We now know that we are the only one running, + * so we finish off the smp_send_stop function. + */ + __cli(); +#ifdef CONFIG_SMP + disable_local_APIC(); +#endif +} + +/* + * Core dump IPI + */ +void smp_crash_funnel_cpu(void) +{ + crash_save_current_state(current); +} diff -urN linux-2.4.7/arch/i386/kernel/head.S linux-2.4.7-mcore/arch/i386/kernel/head.S --- linux-2.4.7/arch/i386/kernel/head.S Wed Jun 20 14:00:53 2001 +++ linux-2.4.7-mcore/arch/i386/kernel/head.S Fri Jul 27 14:58:14 2001 @@ -43,6 +43,18 @@ */ startup_32: /* + * GDT is invalid if we're booted by bootimg, so reload it now + */ + lgdt %cs:_gdt_descr-__PAGE_OFFSET + ljmp $(__KERNEL_CS),$1f-__PAGE_OFFSET + +_gdt_descr: + .word 31 + .long SYMBOL_NAME(gdt_table)-__PAGE_OFFSET + +1: + +/* * Set segments to known values */ cld diff -urN linux-2.4.7/arch/i386/kernel/process.c linux-2.4.7-mcore/arch/i386/kernel/process.c --- linux-2.4.7/arch/i386/kernel/process.c Wed Jul 4 17:41:33 2001 +++ linux-2.4.7-mcore/arch/i386/kernel/process.c Fri Jul 27 14:58:14 2001 @@ -46,6 +46,9 @@ #ifdef CONFIG_MATH_EMULATION #include #endif +#ifdef CONFIG_BOOTIMG +#include +#endif #include @@ -345,6 +348,19 @@ void machine_restart(char * __unused) { +#ifdef CONFIG_MCL_COREDUMP + extern char *panicmsg; + /* + * Only call bootimg if we have a valid descriptor and + * we are in a panic() context. + */ + if (panicmsg) +#endif +#ifdef CONFIG_BOOTIMG + if (bootimg_dsc.page_dir) + boot_image(); +#endif + #if CONFIG_SMP /* * Stop all CPUs and turn off local APICs and the IO-APIC, so diff -urN linux-2.4.7/arch/i386/kernel/setup.c linux-2.4.7-mcore/arch/i386/kernel/setup.c --- linux-2.4.7/arch/i386/kernel/setup.c Wed Jul 11 12:31:44 2001 +++ linux-2.4.7-mcore/arch/i386/kernel/setup.c Fri Jul 27 14:58:14 2001 @@ -104,6 +104,9 @@ #include #include #include +#ifdef CONFIG_MCL_COREDUMP +#include +#endif /* * Machine setup.. */ @@ -768,6 +771,7 @@ void __init setup_arch(char **cmdline_p) { unsigned long bootmap_size, low_mem_size; + unsigned long bootmap_pages, crash_pages; unsigned long start_pfn, max_pfn, max_low_pfn; int i; @@ -876,6 +880,21 @@ pages_to_mb(highend_pfn - highstart_pfn)); } #endif + +#ifdef CONFIG_MCL_COREDUMP + bootmap_pages = bootmem_bootmap_pages(max_low_pfn); + crash_pages = crash_pages_needed(); + + printk("start_pfn: %d, bootmap_pages: %d\n", start_pfn, bootmap_pages); + + crash_init((u_long)phys_to_virt(PFN_PHYS(start_pfn)), + (u_long)phys_to_virt(PFN_PHYS(LOW_OFFSET + start_pfn)), + (u_long)phys_to_virt(PFN_PHYS(LOW_OFFSET + start_pfn + + crash_pages))); + + printk("new start_pfn: %08lx\n", PFN_PHYS(start_pfn)); + printk("crash map starts at %lx\n",(start_pfn+bootmap_pages)*PAGE_SIZE); +#endif /* * Initialize the boot-time allocator (with low memory only): */ @@ -921,8 +940,8 @@ * the (very unlikely) case of us accidentally initializing the * bootmem allocator with an invalid RAM area. */ - reserve_bootmem(HIGH_MEMORY, (PFN_PHYS(start_pfn) + - bootmap_size + PAGE_SIZE-1) - (HIGH_MEMORY)); + reserve_bootmem(HIGH_MEMORY, (PFN_PHYS(start_pfn) + bootmap_size + + ((1+crash_pages)*PAGE_SIZE) + PAGE_SIZE-1) - (HIGH_MEMORY)); /* * reserve physical page 0 - it's a special BIOS page on many boxes, @@ -930,6 +949,16 @@ */ reserve_bootmem(0, PAGE_SIZE); +#ifdef CONFIG_BOOTIMG + /* + * bootimg(8) reads the old parameter block. Note that the copy in + * empty_zero_page will vanish when mem_init runs. (Should we + * memcpy(phys_to_virt(0x90000), PARAM, PAGE_SIZE); + * now ?) + */ + reserve_bootmem(0x90000, PAGE_SIZE); +#endif + #ifdef CONFIG_SMP /* * But first pinch a few for the stack/trampoline stuff @@ -947,6 +976,12 @@ find_smp_config(); #endif paging_init(); +#ifdef CONFIG_MCL_COREDUMP + /* + * Reserve crash pages. + */ + crash_mark_dump_reserved(); +#endif #ifdef CONFIG_X86_IO_APIC /* * get boot-time SMP configuration: @@ -959,6 +994,7 @@ #endif #ifdef CONFIG_BLK_DEV_INITRD + printk("caution: initrd may overwrite dump\n"); /* phro */ if (LOADER_TYPE && INITRD_START) { if (INITRD_START + INITRD_SIZE <= (max_low_pfn << PAGE_SHIFT)) { reserve_bootmem(INITRD_START, INITRD_SIZE); diff -urN linux-2.4.7/arch/i386/kernel/smp.c linux-2.4.7-mcore/arch/i386/kernel/smp.c --- linux-2.4.7/arch/i386/kernel/smp.c Tue Feb 13 17:13:43 2001 +++ linux-2.4.7-mcore/arch/i386/kernel/smp.c Fri Jul 27 14:58:14 2001 @@ -21,6 +21,9 @@ #include #include +#ifdef CONFIG_MCL_COREDUMP +#include +#endif /* * Some notes on x86 processor bugs affecting SMP operation: * @@ -483,7 +486,7 @@ return 0; } -static void stop_this_cpu (void * dummy) +void stop_this_cpu (void * dummy) { /* * Remove this CPU: diff -urN linux-2.4.7/arch/i386/kernel/traps.c linux-2.4.7-mcore/arch/i386/kernel/traps.c --- linux-2.4.7/arch/i386/kernel/traps.c Wed Jun 20 16:59:44 2001 +++ linux-2.4.7-mcore/arch/i386/kernel/traps.c Fri Jul 27 14:58:14 2001 @@ -49,6 +49,10 @@ #include +#ifdef CONFIG_MCL_COREDUMP +#include +#endif + asmlinkage int system_call(void); asmlinkage void lcall7(void); asmlinkage void lcall27(void); @@ -222,7 +226,11 @@ show_registers(regs); spin_unlock_irq(&die_lock); - do_exit(SIGSEGV); +#ifdef CONFIG_MCL_COREDUMP + if(panic_on_oops) + panic("die"); +#endif + do_exit(SIGSEGV);/* NOTREACHED */ } static inline void die_if_kernel(const char * str, struct pt_regs * regs, long err) diff -urN linux-2.4.7/arch/i386/vmlinux.lds linux-2.4.7-mcore/arch/i386/vmlinux.lds --- linux-2.4.7/arch/i386/vmlinux.lds Mon Jul 2 17:40:14 2001 +++ linux-2.4.7-mcore/arch/i386/vmlinux.lds Fri Jul 27 14:58:14 2001 @@ -20,6 +20,13 @@ .rodata : { *(.rodata) *(.rodata.*) } .kstrtab : { *(.kstrtab) } + . = ALIGN(16); /* Relocatable bootimage code */ + __bootimg_start = .; + .bootimg : { + *(.bootimg) + } + __bootimg_end = .; + . = ALIGN(16); /* Exception table */ __start___ex_table = .; __ex_table : { *(__ex_table) } diff -urN linux-2.4.7/arch/ppc/config.in linux-2.4.7-mcore/arch/ppc/config.in --- linux-2.4.7/arch/ppc/config.in Wed Jul 18 10:14:01 2001 +++ linux-2.4.7-mcore/arch/ppc/config.in Fri Jul 27 14:58:14 2001 @@ -376,4 +376,5 @@ bool 'Magic SysRq key' CONFIG_MAGIC_SYSRQ bool 'Include kgdb kernel debugger' CONFIG_KGDB bool 'Include xmon kernel debugger' CONFIG_XMON +bool 'Kernel Core Dump Facility' CONFIG_MCL_COREDUMP endmenu diff -urN linux-2.4.7/arch/ppc/kernel/crash.c linux-2.4.7-mcore/arch/ppc/kernel/crash.c --- linux-2.4.7/arch/ppc/kernel/crash.c Wed Dec 31 19:00:00 1969 +++ linux-2.4.7-mcore/arch/ppc/kernel/crash.c Fri Jul 27 14:58:14 2001 @@ -0,0 +1,93 @@ +/* + * linux/arch/ppc/crash.c + * + * Architecture dependant code for MCL in-memory core dump. + */ +#include +#include +#include +#include +#include + +unsigned long crash_console_sp = 0; +extern unsigned long cpu_present_mask; + +inline void crash_save_regs(void) { + static unsigned long gprs[32]; + + __asm__ __volatile__("stw %%r0,%0" : "=m"(gprs[0])); + __asm__ __volatile__("stw %%r1,%0" : "=m"(gprs[1])); + __asm__ __volatile__("stw %%r2,%0" : "=m"(gprs[2])); + __asm__ __volatile__("stw %%r3,%0" : "=m"(gprs[3])); + __asm__ __volatile__("stw %%r4,%0" : "=m"(gprs[4])); + __asm__ __volatile__("stw %%r5,%0" : "=m"(gprs[5])); + __asm__ __volatile__("stw %%r6,%0" : "=m"(gprs[6])); + __asm__ __volatile__("stw %%r7,%0" : "=m"(gprs[7])); + __asm__ __volatile__("stw %%r8,%0" : "=m"(gprs[8])); + __asm__ __volatile__("stw %%r9,%0" : "=m"(gprs[9])); + __asm__ __volatile__("stw %%r10,%0" : "=m"(gprs[10])); + __asm__ __volatile__("stw %%r11,%0" : "=m"(gprs[11])); + __asm__ __volatile__("stw %%r12,%0" : "=m"(gprs[12])); + __asm__ __volatile__("stw %%r13,%0" : "=m"(gprs[13])); + __asm__ __volatile__("stw %%r14,%0" : "=m"(gprs[14])); + __asm__ __volatile__("stw %%r16,%0" : "=m"(gprs[16])); + __asm__ __volatile__("stw %%r17,%0" : "=m"(gprs[17])); + __asm__ __volatile__("stw %%r18,%0" : "=m"(gprs[18])); + __asm__ __volatile__("stw %%r19,%0" : "=m"(gprs[19])); + __asm__ __volatile__("stw %%r20,%0" : "=m"(gprs[20])); + __asm__ __volatile__("stw %%r21,%0" : "=m"(gprs[21])); + __asm__ __volatile__("stw %%r22,%0" : "=m"(gprs[22])); + __asm__ __volatile__("stw %%r23,%0" : "=m"(gprs[23])); + __asm__ __volatile__("stw %%r24,%0" : "=m"(gprs[24])); + __asm__ __volatile__("stw %%r25,%0" : "=m"(gprs[25])); + __asm__ __volatile__("stw %%r26,%0" : "=m"(gprs[26])); + __asm__ __volatile__("stw %%r27,%0" : "=m"(gprs[27])); + __asm__ __volatile__("stw %%r28,%0" : "=m"(gprs[28])); + __asm__ __volatile__("stw %%r29,%0" : "=m"(gprs[29])); + __asm__ __volatile__("stw %%r30,%0" : "=m"(gprs[30])); + __asm__ __volatile__("stw %%r31,%0" : "=m"(gprs[31])); + + panic_regs = regs; +} + +/* + * Save the current stack pointer. + */ +void crash_save_current_state(struct task_struct *tp) +{ + register unsigned long sp __asm__("%1"); + tp->tss.ksp = sp; + panic_ksp[smp_processor_id()] = sp; + mb(); + + save_core(); + + crash_halt_or_reboot(1); +} + +/* + * If we are not the panicking thread, we simply halt. Otherwise, + * we take care of calling the reboot code. + */ +void crash_halt_or_reboot(int boot_cpu) +{ + printk("restarting system.\n"); + machine_restart(NULL); +} + +void crash_cleanup_smp_state(void) +{ + /* + * Not much to be done here for ppc. + */ + __cli(); +} + + +/* + * Crash IPI + */ +void smp_crash_funnel_cpu(void) +{ + crash_save_current_state(current); +} diff -urN linux-2.4.7/arch/ppc/mm/init.c linux-2.4.7-mcore/arch/ppc/mm/init.c --- linux-2.4.7/arch/ppc/mm/init.c Mon Jul 2 17:34:57 2001 +++ linux-2.4.7-mcore/arch/ppc/mm/init.c Fri Jul 27 14:58:14 2001 @@ -1000,7 +1000,7 @@ */ void __init do_init_bootmem(void) { - unsigned long start, size; + unsigned long start, size, crash_pages = 0; int i; /* @@ -1010,6 +1010,11 @@ * for 4GB of memory, using 4kB pages), plus 1 page * (in case the address isn't page-aligned). */ + +#ifdef CONFIG_MCL_COREDUMP + crash_pages = crash_pages_needed(); +#endif + start = 0; size = 0; for (i = 0; i < phys_avail.n_regions; ++i) { @@ -1019,7 +1024,7 @@ continue; start = a; size = s; - if (s >= 33 * PAGE_SIZE) + if (s >= (33 + crash_pages) * PAGE_SIZE) break; } start = PAGE_ALIGN(start); @@ -1027,13 +1032,24 @@ boot_mapsize = init_bootmem(start >> PAGE_SHIFT, total_lowmem >> PAGE_SHIFT); - /* remove the bootmem bitmap from the available memory */ - mem_pieces_remove(&phys_avail, start, boot_mapsize, 1); +#ifdef CONFIG_MCL_COREDUMP + crash_init((u_long)phys_to_virt(PFN_PHYS(start)), + (u_long)phys_to_virt(PFN_PHYS(start + 33)), + (u_long)phys_to_virt(PFN_PHYS(start + 33 + crash_pages))); +#endif + + /* remove the bootmem bitmap and crash memory from the available memory */ + mem_pieces_remove(&phys_avail, start, + boot_mapsize + crash_pages * PAGE_SIZE, 1); /* add everything in phys_avail into the bootmem map */ for (i = 0; i < phys_avail.n_regions; ++i) free_bootmem(phys_avail.regions[i].address, phys_avail.regions[i].size); + +#ifdef CONFIG_MCL_COREDUMP + crash_mark_dump_reserved(); +#endif init_bootmem_done = 1; } diff -urN linux-2.4.7/drivers/char/misc.c linux-2.4.7-mcore/drivers/char/misc.c --- linux-2.4.7/drivers/char/misc.c Thu May 24 18:14:08 2001 +++ linux-2.4.7-mcore/drivers/char/misc.c Fri Jul 27 14:58:14 2001 @@ -76,6 +76,8 @@ extern int pmu_device_init(void); extern int tosh_init(void); +extern int crash_init_chrdev(void); + static int misc_read_proc(char *buf, char **start, off_t offset, int len, int *eof, void *private) { @@ -252,6 +254,9 @@ int __init misc_init(void) { create_proc_read_entry("misc", 0, 0, misc_read_proc, NULL); +#ifdef CONFIG_MCL_COREDUMP + crash_init_chrdev(); +#endif #ifdef CONFIG_MVME16x rtc_MK48T08_init(); #endif diff -urN linux-2.4.7/drivers/char/sysrq.c linux-2.4.7-mcore/drivers/char/sysrq.c --- linux-2.4.7/drivers/char/sysrq.c Fri Feb 9 14:30:22 2001 +++ linux-2.4.7-mcore/drivers/char/sysrq.c Fri Jul 27 14:58:14 2001 @@ -90,6 +90,9 @@ printk("Resetting\n"); machine_restart(NULL); break; + case 'c': + panic("sysrq"); /* C -- crash and save core */ + break; case 'o': /* O -- power off */ if (sysrq_power_off) { printk("Power off\n"); @@ -148,7 +151,7 @@ printk("Boot "); if (sysrq_power_off) printk("Off "); - printk("Sync Unmount showPc showTasks showMem loglevel0-8 tErm kIll killalL\n"); + printk("Sync Unmount showPc showTasks showMem loglevel0-8 tErm kIll killalL Crash\n"); /* Don't use 'A' as it's handled specially on the Sparc */ } diff -urN linux-2.4.7/include/asm-alpha/crash.h linux-2.4.7-mcore/include/asm-alpha/crash.h --- linux-2.4.7/include/asm-alpha/crash.h Wed Dec 31 19:00:00 1969 +++ linux-2.4.7-mcore/include/asm-alpha/crash.h Fri Jul 27 14:58:14 2001 @@ -0,0 +1,18 @@ +#ifndef __ASM_CRASH_H +#define __ASM_CRASH_H + +#include + +/* + * The alpha's console eats bits toward the end of memory. Stay clear. + */ +#define UPPER_MEM_BACKUP 1000 +#define LOWER_MEM_FORWARD 0 + +extern unsigned int console_crash; +extern unsigned long crash_console_sp; + +void __crash_callin(void); +void crash_callin(void); + +#endif diff -urN linux-2.4.7/include/asm-i386/bootimg.h linux-2.4.7-mcore/include/asm-i386/bootimg.h --- linux-2.4.7/include/asm-i386/bootimg.h Wed Dec 31 19:00:00 1969 +++ linux-2.4.7-mcore/include/asm-i386/bootimg.h Fri Jul 27 15:08:49 2001 @@ -0,0 +1,128 @@ +/* asm-i386/bootimg.h - Boot image, i386-specific code */ + +/* Written 2000 by Werner Almesberger */ + +/* + * When porting bootimg(2) to a new architcture, you need to adapt the + * functions and definitions in this file. + */ + + +#ifndef _ASM_I386_BOOTIMG_H +#define _ASM_I386_BOOTIMG_H + +#include +#include + +#ifdef CONFIG_SMP +#include +#include +#endif + + +/* + * The memory page with the code currently executing has been copied from + * old_page to new_page. Jump there. + * + * Note: flush_icache_range has already been called on the new page. + */ + +static inline void jump_relocated(unsigned long old_page,unsigned long new_page) +{ + int tmp; + + __asm__ __volatile__( + "stc\n\t" + "call 1f\n" + "1:\tjnc 2f\n\t" + "popl %0\n\t" + "addl %1,%0\n\t" + "addl %1,%%esp\n\t" + "clc\n\t" + "jmp *%0\n" + "2:" + : "=&r" (tmp) : "r" (new_page-old_page)); +} + + +/* + * Stop paging, such that + * - page tables can be overwritten + * - all physical memory can be accessed + * - all physical memory is identity-mapped + * + * (Other rules are possible, but need to be encoded in bootimg(8).) + */ + +static inline void stop_paging(void) +{ + unsigned long msw; + + __asm__ __volatile__( + "movl %%cr0,%0\n\t" + "andl $0x7fffffff,%0\n\t" + "movl %0,%%cr0\n\t" + "jmp 1f\n\t" /* i486 and such */ + "1:" + : "=&r" (msw) : : "memory"); +} + + +/* + * Stop any remaining concurrency in the system. If become_only_thread fails + * but the system is still usable, become_only_thread should return an error + * code. If no recovery is possible, it may as well panic. + */ + +static inline int become_only_thread(void) +{ +#ifdef CONFIG_SMP + smp_send_stop(); + disable_IO_APIC(); +#endif + cli(); + return 0; +} + + +/* + * A conservative estimate of the number of bytes relocate_and_jump allocated + * on the stack. This is only used for sanity checking before running code, + * because we can't recover from failure in relocate_and_jump. + */ + +#define RESERVE_MIN_RELOC_STACK 256 + + +/* + * Change the stack pointer such that stack is at the end of the specified + * page. No data on the old stack will be accessed anymore, so no copying is + * required. + */ + +static inline void stack_on_page(void *page) +{ + __asm__ __volatile__( + "push %%ds\n\t" + "pop %%ss\n\t" + "movl %0,%%esp\n\t" + "addl $0x1000,%%esp\n\t" + : : "r" (page)); +} + +/* + * Set up things such that the kernel will be comfortable (e.g. some + * architectures expect the boot loader to set registers in certain ways), + * and then jump to the kernel's entry address. + */ + +static inline void jump_to_kernel(void (*kernel_entry)(void)) +{ + __asm__ __volatile__( + "mov $0x90000,%%esi\n\t" + : : ); + + kernel_entry(); +} + +#endif diff -urN linux-2.4.7/include/asm-i386/crash.h linux-2.4.7-mcore/include/asm-i386/crash.h --- linux-2.4.7/include/asm-i386/crash.h Wed Dec 31 19:00:00 1969 +++ linux-2.4.7-mcore/include/asm-i386/crash.h Fri Jul 27 14:58:14 2001 @@ -0,0 +1,15 @@ +#ifndef __ASM_CRASH_H +#define __ASM_CRASH_H + +#define UPPER_MEM_BACKUP 0 +#define LOWER_MEM_FORWARD 0 +#define LOW_OFFSET 100 + +/* + * These two functions are inlined on alpha. That's why they appear + * in the arch dependent include file. + */ +void crash_save_current_state(struct task_struct *); +void crash_halt_or_reboot(int); + +#endif diff -urN linux-2.4.7/include/asm-ppc/crash.h linux-2.4.7-mcore/include/asm-ppc/crash.h --- linux-2.4.7/include/asm-ppc/crash.h Wed Dec 31 19:00:00 1969 +++ linux-2.4.7-mcore/include/asm-ppc/crash.h Fri Jul 27 14:58:14 2001 @@ -0,0 +1,18 @@ +#ifndef __ASM_CRASH_H +#define __ASM_CRASH_H + +/* OF Eats memory but seems to stay clear from 0xc1700000-0xc7b2a004 */ +#define UPPER_MEM_BACKUP 1500 +#define LOWER_MEM_FORWARD 6000 + +/* + * These two functions are inlined on alpha. That's why they appear + * in the arch dependent include file. + */ +void crash_save_current_state(struct task_struct *); +void crash_halt_or_reboot(int); + +extern unsigned int console_crash; +extern unsigned long crash_console_sp; + +#endif diff -urN linux-2.4.7/include/linux/bootimg.h linux-2.4.7-mcore/include/linux/bootimg.h --- linux-2.4.7/include/linux/bootimg.h Wed Dec 31 19:00:00 1969 +++ linux-2.4.7-mcore/include/linux/bootimg.h Fri Jul 27 14:58:14 2001 @@ -0,0 +1,84 @@ +/* linux/bootimg.h - Boot image, general definitions */ + +/* Written 2000 by Werner Almesberger */ + + +#ifndef _LINUX_BOOTIMG_H +#define _LINUX_BOOTIMG_H + + +/* + * Constraints on image_map: + * - each image_map[n] is the virtual address of a page-sized memory region + * readable by the user + * - currently, image_map[n] is not required to be page-aligned, but this may + * change in the future if we want to map pages directly to lower memory + * pressure (NB: mapping works for ELF and plain binary images, but usually + * not for (b)zImages, because the prepended boot and setup sectors + * mis-align them) + * + * Constraints on load_map: + * - each load_map[] is the physical address of a page in RAM + */ + +struct boot_image { + void **image_map; /* pointers to image pages in user memory */ + int pages; /* length in pages */ + unsigned long *load_map;/* list of destination pages (physical addr) */ + unsigned long start; /* jump to this physical address */ + int flags; /* for future use, must be zero for now */ +}; + + +#ifdef __KERNEL__ + +#define __bootimg __attribute__ ((__section__ (".bootimg"))) + + +struct bootimg_dsc { + unsigned long self; /* code page ALL ADDRESSES */ + unsigned long scratch; /* scratch page ARE PHYSICAL !*/ + unsigned long **page_dir; /* src & dst page tables */ + void (*jump_to)(void); /* start address */ + int pages; /* number of pages */ + unsigned long csum; /* Kernel Image checksum */ +}; + +/* + * page_dir contains pointers to pages containing pointers to pages. We call + * page_dir a "directory" and the page page_dir[n] points to a "table". The + * first PAGES_PER_TABLE/2 entries of page_dir are for source pages, and other + * half are for destination pages. + */ + +/* + * Note that the definitions used here do not necessarily correspond to the + * architecture-specific PTRS_PER_PTE, __pte_offset, etc. + */ + +#define PAGES_PER_TABLE (PAGE_SIZE/sizeof(void *)) +#define FROM_TABLE(i) ((i)/PAGES_PER_TABLE) +#define TO_TABLE(i) ((i)/PAGES_PER_TABLE+PAGES_PER_TABLE/2) +#define PAGE_NR(i) ((i) % PAGES_PER_TABLE) + + +extern char __bootimg_start,__bootimg_end; /* linker segment boundaries */ +extern unsigned long *unity_page; /* unity-mapped page for i386 */ + +/* + * relocate_and_jump runs in its own page with its own stack. This makes it + * difficult to pass parameters. The solution chosen here is to use the global + * variable bootimg_dsc, which is copied into an "auto" variable by + * relocate_and_jump before any copying or relocation takes place. + */ + +extern struct bootimg_dsc bootimg_dsc; + +typedef void (*relocate_and_jump_t)(void); + +void relocate_and_jump(void); +int boot_image(void); + +#endif /* __KERNEL__ */ + +#endif diff -urN linux-2.4.7/include/linux/crash.h linux-2.4.7-mcore/include/linux/crash.h --- linux-2.4.7/include/linux/crash.h Wed Dec 31 19:00:00 1969 +++ linux-2.4.7-mcore/include/linux/crash.h Fri Jul 27 14:58:14 2001 @@ -0,0 +1,119 @@ +#ifndef __LINUX_CRASH_H +#define __LINUX_CRASH_H + +/* defines for interfacing with user-space (ioctls, etc) */ +struct ioctl_getdump { + unsigned long kva; + unsigned long buf; +}; + +#define CRASH_IOC_MAGIC 'C' + +#define CRASH_IOCFREEDUMP _IO(CRASH_IOC_MAGIC, 0) +#define CRASH_IOCGETDUMP _IOWR(CRASH_IOC_MAGIC, 1, struct ioctl_getdump) +#define CRASH_IOCBOOTIMG _IOWR(CRASH_IOC_MAGIC, 2, struct boot_image) +#define CRASH_IOCVERSION _IO(CRASH_IOC_MAGIC, 3) + +/* kernel-only part of crash.h */ +#ifdef __KERNEL__ +#include + +#define CRASH_K_MINOR (1) +#define CRASH_K_MAJOR (0) + +/* + * Crash prototypes. + */ +void save_core(void); +void crash_mark_dump_reserved(void); +void crash_init(u_long bootmap_va, u_long crash_va, u_long end_alloc_va); +u_long crash_pages_needed(void); +void smp_crash_funnel_cpu(void); +void crash_cleanup_smp_state(void); + +/* + * Arch dependant crash.c funcs + */ +void crash_save_current_state(struct task_struct *); +void crash_halt_or_reboot(int); +inline void crash_save_regs(void); + +/* + * Crash globals + */ +extern u_long crash_dump_header; +extern volatile u_long panic_ksp[]; +extern volatile int crash_release; +extern int panic_on_oops; +extern char *panicmsg; +extern int panic_processor; +extern int crash_perform_sync; +extern unsigned long *panic_regs; + +/* + * symbols not exported by linux header files + */ +extern void stop_this_cpu(void *); + +/* struct crash_map_hdr located at byte offset 0 */ +/* on-disk formats */ + +#define trunc_page(x) ((void *)(((unsigned long)(x)) & ~((unsigned long)(PAGE_SIZE - 1)))) +#define round_page(x) trunc_page(((unsigned long)(x)) + ((unsigned long)(PAGE_SIZE - 1))) + +#define CRASH_MAGIC 0x9a8bccdd +#define CRASH_SOURCE_PAGES 128 +#define CRASH_SUB_MAP_BYTES ((u_long)round_page((CRASH_SOURCE_PAGES+1)*sizeof(u_long))) +#define CRASH_SUB_MAP_PAGES (CRASH_SUB_MAP_BYTES / PAGE_SIZE) +#define CRASH_UNCOMPR_BUF_PAGES (CRASH_SOURCE_PAGES + CRASH_SUB_MAP_PAGES) +#define CRASH_COMPR_BUF_PAGES (CRASH_UNCOMPR_BUF_PAGES + (CRASH_UNCOMPR_BUF_PAGES/4)) +#define CRASH_COMPESS_PRIME_PAGES (2*CRASH_COMPR_BUF_PAGES) +#define CRASH_ZALLOC_PAGES 16*5*2 /* 2 to handle crash in crash */ +#define CRASH_LOW_WATER_PAGES 100 + +#define CRASH_CPU_TIMEOUT 5000 /* 5 sec wait for other cpus to stop */ + +#define CRASH_MARK_RESERVED(addr) (set_bit(PG_reserved,&mem_map[MAP_NR(addr)].flags)) +#define CRASH_CLEAR_RESERVED(addr) (clear_bit(PG_reserved,&mem_map[MAP_NR(addr)].flags)) +#define CRASH_MARK_BOOT_RESERVED(addr) reserve_bootmem(virt_to_phys((void *)addr), PAGE_SIZE); + +typedef int boolean_t; + +#define TRUE 1 +#define FALSE 0 + +/* mem structure */ +struct mem_crash_map_hdr { + long magic[4]; /* identify crash dump */ + u_long map; /* location of map */ + u_long map_pages; + u_long data_pages; + u_long compr_units; + u_long boot_reserved_start; + u_long boot_reserved_end; +}; +struct mem_crash_map_entry { + u_long src_va; /* source start of larger non-contig + * block. a src_va of -1 means that + * the dest_page_va is the location of + * the next map page */ + u_long dest_page_va; /* dest of this sub block */ + u_long check_sum; /* check_sum for dest data */ +}; + +/* file structure */ +struct crash_map_hdr { + long magic[4]; /* identify crash dump */ + int blk_size; /* block size for this device */ + int map_block; /* location of map */ + int map_blocks; /* number of blocks for map */ +}; +struct crash_map_entry { + u_long start_va; /* virtual address */ + char *exp_data; /* expanded data in memory */ + int start_blk; /* device location */ + int num_blks; +}; + +#endif /* __KERNEL__ */ +#endif /* __LINUX_CRASH_H */ diff -urN linux-2.4.7/include/linux/mm.h linux-2.4.7-mcore/include/linux/mm.h --- linux-2.4.7/include/linux/mm.h Fri Jul 20 15:52:18 2001 +++ linux-2.4.7-mcore/include/linux/mm.h Fri Jul 27 15:08:49 2001 @@ -282,6 +282,9 @@ #define PG_inactive_clean 11 #define PG_highmem 12 #define PG_checked 13 /* kill me in 2.5.. */ +#define PG_free 14 +#define PG_anon 15 +#define PG_shm 16 /* bits 21-29 unused */ #define PG_arch_1 30 #define PG_reserved 31 @@ -333,6 +336,9 @@ #define PageSlab(page) test_bit(PG_slab, &(page)->flags) #define PageSwapCache(page) test_bit(PG_swap_cache, &(page)->flags) #define PageReserved(page) test_bit(PG_reserved, &(page)->flags) +#define PageFree(page) (test_bit(PG_free, &(page)->flags)) +#define PageAnon(page) (test_bit(PG_anon, &(page)->flags)) +#define PageShm(page) (test_bit(PG_shm, &(page)->flags)) #define PageSetSlab(page) set_bit(PG_slab, &(page)->flags) #define PageSetSwapCache(page) set_bit(PG_swap_cache, &(page)->flags) diff -urN linux-2.4.7/include/linux/reboot.h linux-2.4.7-mcore/include/linux/reboot.h --- linux-2.4.7/include/linux/reboot.h Fri Feb 9 17:46:13 2001 +++ linux-2.4.7-mcore/include/linux/reboot.h Fri Jul 27 14:58:14 2001 @@ -20,6 +20,7 @@ * CAD_OFF Ctrl-Alt-Del sequence sends SIGINT to init task. * POWER_OFF Stop OS and remove all power from system, if possible. * RESTART2 Restart system using given command string. + * COREDUMP We're taking a core dump, secondary cpus already stopped. */ #define LINUX_REBOOT_CMD_RESTART 0x01234567 @@ -28,7 +29,7 @@ #define LINUX_REBOOT_CMD_CAD_OFF 0x00000000 #define LINUX_REBOOT_CMD_POWER_OFF 0x4321FEDC #define LINUX_REBOOT_CMD_RESTART2 0xA1B2C3D4 - +#define LINUX_REBOOT_CMD_COREDUMP 0x9A8BCCDD #ifdef __KERNEL__ diff -urN linux-2.4.7/include/linux/sysctl.h linux-2.4.7-mcore/include/linux/sysctl.h --- linux-2.4.7/include/linux/sysctl.h Fri Jul 20 15:52:18 2001 +++ linux-2.4.7-mcore/include/linux/sysctl.h Fri Jul 27 15:08:44 2001 @@ -118,7 +118,8 @@ KERN_SHMPATH=48, /* string: path to shm fs */ KERN_HOTPLUG=49, /* string: path to hotplug policy agent */ KERN_IEEE_EMULATION_WARNINGS=50, /* int: unimplemented ieee instructions */ - KERN_S390_USER_DEBUG_LOGGING=51 /* int: dumps of user faults */ + KERN_S390_USER_DEBUG_LOGGING=51, /* int: dumps of user faults */ + KERN_PANIC_ON_OOPS=52 /* int: panic on oops enable */ }; diff -urN linux-2.4.7/init/main.c linux-2.4.7-mcore/init/main.c --- linux-2.4.7/init/main.c Thu Jul 5 14:31:58 2001 +++ linux-2.4.7-mcore/init/main.c Fri Jul 27 15:53:35 2001 @@ -70,6 +70,10 @@ #include #endif +#ifdef CONFIG_BOOTIMG +#include +#endif + /* * Versions of gcc older than that listed below may actually compile * and link okay, but the end product can have subtle run time bugs. @@ -526,10 +530,14 @@ char * command_line; unsigned long mempages; extern char saved_command_line[]; +#if defined(CONFIG_BOOTIMG) && defined(CONFIG_X86_LOCAL_APIC) + unsigned long value; +#endif /* * Interrupts are still disabled. Do necessary setups, then * enable them */ + printk("start_kernel\n"); lock_kernel(); printk(linux_banner); setup_arch(&command_line); @@ -547,6 +555,11 @@ * this. But we do want output early, in case something goes wrong. */ console_init(); + +#ifdef CONFIG_BOOTIMG + unity_page = alloc_bootmem_pages(PAGE_SIZE); + printk("unity_page addr: %p\n",unity_page); +#endif #ifdef CONFIG_MODULES init_modules(); #endif @@ -562,6 +575,15 @@ kmem_cache_init(); sti(); +#if defined(CONFIG_BOOTIMG) && defined(CONFIG_X86_LOCAL_APIC) + /* If we don't make sure the APIC is enabled, AND the LVT0 + * register is programmed properly, we won't get timer interrupts + */ + setup_local_APIC(); + + value = apic_read(APIC_LVT0); + apic_write_around(APIC_LVT0, value & ~APIC_LVT_MASKED); +#endif calibrate_delay(); #ifdef CONFIG_BLK_DEV_INITRD if (initrd_start && !initrd_below_start_ok && diff -urN linux-2.4.7/kernel/Makefile linux-2.4.7-mcore/kernel/Makefile --- linux-2.4.7/kernel/Makefile Fri Dec 29 17:07:24 2000 +++ linux-2.4.7-mcore/kernel/Makefile Fri Jul 27 14:58:14 2001 @@ -19,6 +19,12 @@ obj-$(CONFIG_UID16) += uid16.o obj-$(CONFIG_MODULES) += ksyms.o obj-$(CONFIG_PM) += pm.o +obj-$(CONFIG_BOOTIMG) += bootimg.o bootimg_pic.o +obj-$(CONFIG_MCL_COREDUMP) += crash.o + +ifneq ($(CONFIG_IRDA_DEFLATE),y) +obj-y += zlib.o +endif ifneq ($(CONFIG_IA64),y) # According to Alan Modra , the -fno-omit-frame-pointer is diff -urN linux-2.4.7/kernel/bootimg.c linux-2.4.7-mcore/kernel/bootimg.c --- linux-2.4.7/kernel/bootimg.c Wed Dec 31 19:00:00 1969 +++ linux-2.4.7-mcore/kernel/bootimg.c Fri Jul 27 15:53:15 2001 @@ -0,0 +1,260 @@ +/* bootimg.c - Boot another (kernel) image */ + +/* Written 2000 by Werner Almesberger */ + + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#if 0 +#define DPRINTK_CONT(format,args...) printk(format,##args) +#else +#define DPRINTK_CONT(format,args...) +#endif +#define DPRINTK(format,args...) DPRINTK_CONT(KERN_DEBUG format,##args) + +unsigned long **bootimg_page_dir; + +struct bootimg_dsc bootimg_dsc; /* communication with PIC */ +unsigned long *unity_page; /* unity-mapped page for i386 */ + +static unsigned long cheese_checksum(unsigned long **page_dir, int num_pages) +{ + int i,csum=0; + + for (i = 0; i < num_pages; i++) + csum += *(unsigned long *)__va(bootimg_page_dir[0][i]); + return csum; +} + +static unsigned long get_identity_mapped_page(void) +{ + set_pgd(pgd_offset(current->active_mm,virt_to_phys(unity_page)), + __pgd((_KERNPG_TABLE + _PAGE_PSE + (virt_to_phys(unity_page)&PGDIR_MASK)))); + return (unsigned long)unity_page; +} + +#if 0 /* Perhaps we'll need this in the future? */ +static void unmap_identity_mapped_page(void) +{ + set_pgd(pgd_offset(current->active_mm,virt_to_phys(unity_page)),__pgd(0)); + __flush_tlb(); +} +#endif + +static int fill_page_dir(unsigned long **page_dir,struct boot_image *image) +{ + int i, count=0; + + memset(bootimg_page_dir,0,PAGE_SIZE); + for (i = 0; i < image->pages; i += PAGES_PER_TABLE) { + unsigned long **table; + int bytes_left; + + table = bootimg_page_dir+FROM_TABLE(i); + *table = (unsigned long *) get_free_page(GFP_KERNEL); + if (!*table) return -ENOMEM; + + memset(*table,0,PAGE_SIZE); + DPRINTK("page %d: from table %p @ %p\n",i,*table,table); + table = bootimg_page_dir+TO_TABLE(i); + *table = (unsigned long *) get_free_page(GFP_KERNEL); + if (!*table) return -ENOMEM; + + bytes_left = (image->pages-i)*sizeof(unsigned long); + if (copy_from_user(*table,image->load_map+i, + bytes_left > PAGE_SIZE ? PAGE_SIZE : bytes_left)) + return -EFAULT; + DPRINTK("page %d: to table %p @ %p\n",i,*table,table); + count+=2; /* 2 pages per loop */ + } + + for (i = 0; i < image->pages; i++) { + unsigned long page = get_free_page(GFP_KERNEL); + void *src; + + if (!page) return -ENOMEM; + count++; + + bootimg_page_dir[FROM_TABLE(i)][PAGE_NR(i)] = + virt_to_phys((void *) page); + if (get_user(src,image->image_map+i) || + copy_from_user((void *) page,src,PAGE_SIZE)) + return -EFAULT; + + DPRINTK("page %d: %p->%p->%p @ %p\n",i,src,(void *) page, + (void *) bootimg_page_dir[FROM_TABLE(i)][PAGE_NR(i)], + &bootimg_page_dir[FROM_TABLE(i)][PAGE_NR(i)]); + } + + DPRINTK("fill_page_dir: %d pages allocated\n", count); + + return 0; +} + + +static void free_page_dir(unsigned long **page_dir) +{ + int i,j,count=0; + + for (i = 0; i < PAGES_PER_TABLE/2; i++) + if (bootimg_page_dir[i]) + for (j = 0; j < PAGES_PER_TABLE; j++) + if (bootimg_page_dir[i][j]) { + free_page((unsigned long) + phys_to_virt(bootimg_page_dir[i][j])); + count++; + } + for (i = 0; i < PAGES_PER_TABLE; i++) + if (bootimg_page_dir[i]) { + free_page((unsigned long) *bootimg_page_dir[i]); + count++; + } + DPRINTK("free_page_dir: %d pages freed\n", count); +} + + +static void convert_table_refs_to_phys(unsigned long **page_dir) +{ + int i; + + DPRINTK("PAGES_PER_TABLE: %d\n",PAGES_PER_TABLE); + for (i = 0; i < PAGES_PER_TABLE; i++) + if (bootimg_page_dir[i]) { + DPRINTK("table %i: mapped %p -> ",i,bootimg_page_dir[i]); + bootimg_page_dir[i] = (unsigned long *) + virt_to_phys(bootimg_page_dir[i]); + DPRINTK_CONT("%p\n",bootimg_page_dir[i]); + } +} + + + +static int fill_bootimg_dsc(struct boot_image *image) +{ + unsigned long scratch; + int error = -ENOMEM; + + if(bootimg_page_dir) { + /* free previously allocated memory */ + free_page_dir(bootimg_page_dir); + free_page((unsigned long) bootimg_page_dir); + DPRINTK("free_page (bootimg_page_dir)\n"); + } + + bootimg_page_dir = (unsigned long **) get_free_page(GFP_KERNEL); + if (!bootimg_page_dir) goto out0; + DPRINTK("get_free_page (bootimg_page_dir)\n"); + + error = fill_page_dir(bootimg_page_dir,image); + if (error) goto out1; + + if(!bootimg_dsc.scratch) { + scratch = get_free_page(GFP_KERNEL); + DPRINTK("get_free_page (scratch)\n"); + } else + scratch = 1; /* already allocated */ + + if (!scratch) goto out1; + /* + * Not all architectures need the code to be identity-mapped, but it + * can't hurt ... + */ + DPRINTK("bootimg_page_dir: mapped %p -> ",bootimg_page_dir); + bootimg_dsc.page_dir = (unsigned long **) virt_to_phys(bootimg_page_dir); + DPRINTK_CONT("%p\n",bootimg_dsc.page_dir); + if(!bootimg_dsc.scratch) + bootimg_dsc.scratch = virt_to_phys((void *) scratch); + bootimg_dsc.jump_to = (void (*)(void)) image->start; + bootimg_dsc.pages = image->pages; + bootimg_dsc.csum = cheese_checksum(bootimg_page_dir, image->pages); + + return 0; + +out1: + free_page_dir(bootimg_page_dir); + free_page((unsigned long) bootimg_page_dir); + DPRINTK("free_page (bootimg_page_dir)\n"); + bootimg_page_dir = 0; +out0: + return error; +} + +extern char *panicmsg; +int boot_image() +{ + relocate_and_jump_t code; + unsigned long code_page; + int error = -ENOMEM; + int i; + + if (cheese_checksum(__va(bootimg_dsc.page_dir),bootimg_dsc.pages) + != bootimg_dsc.csum) + printk("Checksum of kernel image failed. Rebooting via BIOS\n"); + + code_page = get_identity_mapped_page(); + if (!code_page) goto out3; + code = (relocate_and_jump_t) virt_to_phys((void *) code_page); + memcpy(code,&__bootimg_start,&__bootimg_end-&__bootimg_start); + + bootimg_dsc.self = code_page; + printk(KERN_INFO "Running boot code at 0x%p\n",code); + + /* + * The point of no return. Not even printk may work after a successful + * return from become_only_thread. + */ + + if (!panicmsg) { + error = become_only_thread(); + if (error) goto out3; + } else { +#ifdef CONFIG_SMP + disable_IO_APIC(); +#endif + __cli(); + } + + convert_table_refs_to_phys((unsigned long **)__va(bootimg_dsc.page_dir)); + stack_on_page(code); + + code(); + + panic("PIC code exec failed"); +out3: + printk("boot_image() failed!\n"); + for(;;); +} + +/* changed from asmlinkage because we're called via an IOCTL on /dev/crash now */ +int sys_bootimg(struct boot_image *user_dsc) +{ + struct boot_image dsc; + + if (!capable(CAP_SYS_ADMIN) || !capable(CAP_SYS_MODULE)) return -EPERM; + if (&__bootimg_end-&__bootimg_start > PAGE_SIZE-RESERVE_MIN_RELOC_STACK) + { + printk(KERN_ERR "boot_image: PIC too large (%d bytes)\n", + &__bootimg_end-&__bootimg_start); + return -EIO; + } + if ((void *) relocate_and_jump != (void *) &__bootimg_start) { + printk(KERN_ERR "boot_image: relocate_and_jump is mis-placed" + "(0x%p != 0x%p)\n",relocate_and_jump,&__bootimg_start); + return -EIO; + } + + if (copy_from_user(&dsc,user_dsc,sizeof(dsc))) return -EFAULT; + if (dsc.pages >= PAGES_PER_TABLE*PAGES_PER_TABLE/2) return -EFBIG; + if (dsc.flags) return -EINVAL; /* for future use */ + return fill_bootimg_dsc(&dsc); +} diff -urN linux-2.4.7/kernel/bootimg_pic.c linux-2.4.7-mcore/kernel/bootimg_pic.c --- linux-2.4.7/kernel/bootimg_pic.c Wed Dec 31 19:00:00 1969 +++ linux-2.4.7-mcore/kernel/bootimg_pic.c Fri Jul 27 14:58:14 2001 @@ -0,0 +1,93 @@ +/* bootimg_pic.c - Boot image, position-independent code */ + +/* Written 2000 by Werner Almesberger */ + +/* + * Strongly inspired by FiPaBoL designed mainly by Otfried Cheong and Roger + * Gammans, and written by the latter. + */ + +/* + * This code is position-independent and must fit in a single page ! + * Furthermore, everything (text+data+stack) has to go into the + * .bootimg segment. + */ + + +#include +#include +#include +#include +#include +#include + +#include + +#define copy_and_swap(from,to) \ + ( { my_copy_page(from,to); \ + tmp = from; \ + from = to; \ + to = tmp; } ) + + +static inline void my_copy_page(unsigned long from,unsigned long to) +{ + unsigned long end = from+PAGE_SIZE; + + do *((unsigned long *) to)++ = *((unsigned long *) from)++; + while (from != end); +} + + +void __bootimg relocate_and_jump(void) +{ + struct bootimg_dsc dsc = bootimg_dsc; + int i; + + stop_paging(); + for (i = 0; i < dsc.pages; i++) { + unsigned long from,to,tmp; + + from = dsc.page_dir[FROM_TABLE(i)][PAGE_NR(i)]; + to = dsc.page_dir[TO_TABLE(i)][PAGE_NR(i)]; + if (from == to) continue; + if (to == dsc.self) { + copy_and_swap(dsc.self,dsc.scratch); + /* WARNING: flush_icache_range MUST BE INLINED !!! */ + flush_icache_range(dsc.self,dsc.self+PAGE_SIZE-1); + jump_relocated(dsc.scratch,dsc.self); + } + else if (to == (unsigned long) dsc.page_dir) + copy_and_swap((unsigned long) dsc.page_dir,dsc.scratch); + else { + /* + * O((n^2-n)/2), sigh ... + */ + unsigned long *table; + int j; + + for (j = i+1; j < dsc.pages; j++) { + table = (unsigned long *) + dsc.page_dir+FROM_TABLE(j); + if (*table == to) { + copy_and_swap(*table,dsc.scratch); + break; + } + if (table[PAGE_NR(j)] == to) { + copy_and_swap(table[PAGE_NR(j)], + dsc.scratch); + break; + } + table = (unsigned long *) + dsc.page_dir+TO_TABLE(j); + if (*table == to) { + copy_and_swap(*table,dsc.scratch); + break; + } + } + } + my_copy_page(from,to); + dsc.scratch = from; + } + jump_to_kernel(dsc.jump_to); +} diff -urN linux-2.4.7/kernel/crash.c linux-2.4.7-mcore/kernel/crash.c --- linux-2.4.7/kernel/crash.c Wed Dec 31 19:00:00 1969 +++ linux-2.4.7-mcore/kernel/crash.c Fri Jul 27 15:53:05 2001 @@ -0,0 +1,882 @@ +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include "../drivers/net/zlib.h" +#include +#include +#include +#include +#include + +#ifdef CONFIG_BOOTIMG +#include +#endif + +static void crash_print_data_around(u_long p); +static void crash_free_page(u_long addr); +static int crash_chksum_page(u_long pg_addr, u_long * sum_addr); +static void *czalloc(void *arg, unsigned int items, unsigned int size); +static void czfree(void *arg, void *ptr); +static u_long crash_alloc_dest_page(void); +static void crash_free_dest_page(u_long dest); +static void init_dest_page_alloc(void); +static int crash_audit_maps(void); +static u_long crash_get_source_page(void); +static u_long crash_update_map(u_long map, u_long src_base, u_long dest, u_long * pages); +static int crash_reset_stream(z_stream * stream); +static boolean_t crash_is_kseg(u_long addr); +static u_long *crash_link(u_long p); +static int crash_chksum(u_long limit, u_long * sum_addr); +static int crash_audit_map_page(u_long map); +static void crash_wait_cpus(void); +static int crash_is_dir_page(struct page *page); + +/* for the /dev/crash interface */ +int crash_init_chrdev(void); +static int crashdev_ioctl(struct inode *, struct file *, unsigned int, unsigned long); + +#define CRASH_DEBUG 1 + +#ifdef CONFIG_BOOTIMG +extern int sys_bootimg(struct boot_image *); +#endif + +static u_long crash_compr_buf; +static u_long crash_uncompr_buf; +static u_long crash_dump_header = 0; +static u_long crash_dest_free_list = 0; +static u_long crash_debug = 0; + +static u_long crash_cur_pfn; + +static u_long src_pages_skipped = 0; +static u_long src_pages_saved = 0; +static u_long dest_pages_free = 0; + +/* this information is saved from within panic() */ +char *panicmsg = (char *)0; +int panic_processor = 0; +int crash_perform_sync = 0; + +u_int console_crash = 0; /* should be moved to alpha branch */ + +typedef struct task_struct *task_t; + +/* + * Threads active at time of panic: + */ +volatile task_t panic_threads[NR_CPUS]; +volatile unsigned long panic_ksp[NR_CPUS]; +unsigned long *panic_regs = NULL; + +int panic_on_oops; /* for /proc/sys/kernel/panic_on_oops */ + +extern unsigned long max_low_pfn; + +u_long crash_zalloc_start, crash_zalloc_end, crash_zalloc_cur; + +/* + * Crash Kernel API functions below + * crash_pages_needed, computes pages needed for header and compression temp + * crash_init, partitions out the allocated pages, sets defaults and + * initializes the character device. + * crash_mark_dump_reserved, marks pages reserved from a previous dump. + * save_core, called at panic time to save a dump to memory. + */ +u_long crash_pages_needed(void) +{ + /* one for the header */ + return (1 + CRASH_ZALLOC_PAGES + CRASH_UNCOMPR_BUF_PAGES + CRASH_COMPR_BUF_PAGES); +} + +void crash_init(u_long bootmap_va, u_long crash_va, u_long end_alloc_va) +{ + struct mem_crash_map_hdr *header; + int i; + + /* the default behavior is not NOT panic on a kernel OOPS */ + panic_on_oops = 0; + + printk("crash_init (crash_va: %08lx)\n", crash_va); + for (i = 0; i < NR_CPUS; i++) + panic_threads[i] = 0; + crash_dump_header = crash_va; + crash_va += PAGE_SIZE; + crash_zalloc_cur = crash_zalloc_start = crash_va; + crash_va += CRASH_ZALLOC_PAGES * PAGE_SIZE; + crash_zalloc_end = crash_uncompr_buf = crash_va; + crash_va += CRASH_UNCOMPR_BUF_PAGES * PAGE_SIZE; + crash_compr_buf = crash_va; + crash_va += CRASH_COMPR_BUF_PAGES * PAGE_SIZE; +#if 0 + if (crash_va != end_alloc_va) + panic("crash_init inconsistency-1\n"); +#endif + + header = (struct mem_crash_map_hdr *)crash_dump_header; +#ifdef CRASH_DEBUG + printk("crash_dump_header %p {\n", header); + printk(" magic[0] = %lx\n", header->magic[0]); + printk(" map = %lx\n", header->map); + printk(" map_pages = %lx\n", header->map_pages); + printk(" data_pages = %lx\n", header->data_pages); + printk(" compr_units = %lx\n", header->compr_units); + printk(" boot_reserved_start = %lx\n", header->boot_reserved_start); + printk(" boot_reserved_end = %lx\n", header->boot_reserved_end); +#endif + + if (header->magic[0] == CRASH_MAGIC) { + printk("crash found\n"); + if ((header->boot_reserved_start != bootmap_va) || + (header->boot_reserved_end != end_alloc_va)) { + /* crash audit will catch the corruption */ + printk("crash_init inconsistency, dump may be corrupted\n"); + } + } else { +printk("memset..."); + memset(header, 0, sizeof(*header)); +printk("done\n"); + } + + header->boot_reserved_start = bootmap_va; + header->boot_reserved_end = end_alloc_va; + +} + +void crash_mark_dump_reserved(void) +{ + struct mem_crash_map_hdr *header; + struct mem_crash_map_entry *m; + + header = (struct mem_crash_map_hdr *)crash_dump_header; + if (header->magic[0] != CRASH_MAGIC) + return; + m = (struct mem_crash_map_entry *)header->map; +#ifdef CRASH_DEBUG + printk("\n\n\ncrash_mark_dump_reserved\n\n"); + printk("crash_dump_header %p {\n", header); + printk(" magic[0] = %lx\n", header->magic[0]); + printk(" map = %lx\n", header->map); + printk(" map_pages = %lx\n", header->map_pages); + printk(" data_pages = %lx\n", header->data_pages); + printk(" compr_units = %lx\n", header->compr_units); + printk(" boot_reserved_start = %lx\n", header->boot_reserved_start); + printk(" boot_reserved_end = %lx\n", header->boot_reserved_end); + printk("mem_crash_map_entry %p {\n", m); + printk(" src_va = %lx\n", m->src_va); + printk(" dest_page_va = %lx\n", m->dest_page_va); + printk(" check_sum = %lx\n", m->check_sum); +#endif + + if (crash_audit_maps()) { + header->magic[0] = 0; + return; + } + + m = (struct mem_crash_map_entry *)header->map; + again: + CRASH_MARK_BOOT_RESERVED(m); + for (; m->src_va; m++) { + if (m->src_va == -1) { + m = (struct mem_crash_map_entry *)m->dest_page_va; + goto again; + } + CRASH_MARK_BOOT_RESERVED(m->dest_page_va); + } + return; +} + +void save_core(void) +{ + int i, j, k; + z_stream stream; + int err; + struct task_struct *tp; + struct mem_crash_map_hdr *header; + u_long *sub_map; + u_long map; + u_long src, dest, unc, cp, src_base, comp_pages; + + k = 0; + dest = 0; + __cli(); + tp = current; + mb(); + if (smp_processor_id() != 0) { /* boot_cpu_id is always 0, i think */ + panic_threads[smp_processor_id()] = tp; + crash_halt_or_reboot(0); + } else { + if (console_crash) + panic_threads[smp_processor_id()] = &init_task_union.task; + else + panic_threads[smp_processor_id()] = tp; + + crash_wait_cpus(); + } + + printk("save_core: started on CPU%d\n", smp_processor_id()); + if (!crash_dump_header) { + printk("save_core: not initialized\n"); + return; + } + + header = (struct mem_crash_map_hdr *)crash_dump_header; + header->magic[0] = 0; + header->map_pages = 0; + header->data_pages = 0; + header->compr_units = 0; + header->map = 0; + + stream.zalloc = czalloc; + stream.zfree = czfree; + stream.opaque = (voidpf) 0; + stream.next_out = (Bytef *) crash_compr_buf; + stream.avail_out = (uInt) (CRASH_COMPR_BUF_PAGES * PAGE_SIZE); + stream.next_in = (Bytef *) crash_uncompr_buf; + stream.avail_in = (uInt) (CRASH_UNCOMPR_BUF_PAGES * PAGE_SIZE); + err = deflateInit(&stream, Z_BEST_SPEED); + if (err != Z_OK) { + printk("save_core: bad return %d from deflateInit\n", err); + return; + } + + init_dest_page_alloc(); + header->map = map = crash_update_map(0, 0, 0, &header->map_pages); + if (!map) { + printk("save_core: no dest pages\n"); + return; + } + crash_cur_pfn = 0; + src_base = 0; + src = 0; + for (;;) { + sub_map = (u_long *) crash_uncompr_buf; + unc = crash_uncompr_buf + CRASH_SUB_MAP_PAGES * PAGE_SIZE; + for (i = 0; i < CRASH_SOURCE_PAGES; i++) { + src = crash_get_source_page(); + if (!src) + break; + if (!i) + src_base = src; + if (!crash_is_kseg(unc) || !crash_is_kseg(src)) { + printk("unc = 0x%lx, src = 0x%lx, i = %d\n", unc, src, i); + i = src = 0; + break; + } + memcpy((void *)unc, (void *)src, PAGE_SIZE); + unc += PAGE_SIZE; + *sub_map++ = src; + } + *sub_map = 0; + if (!i && !src) + break; + err = deflate(&stream, Z_FINISH); + if (!(err == Z_STREAM_END)) { + deflateEnd(&stream); + printk("save_core: bad return %d from deflate, src_base = 0x%lx\n", err, + src_base); + return; + } + comp_pages = (u_long) round_page(stream.total_out) / PAGE_SIZE; + if (crash_debug) + printk("src_base = 0x%lx compressed data in 0x%lx pages\n", src_base, + comp_pages); + + cp = crash_compr_buf; + j = 0; + if (crash_debug) + printk("\nsrc = %lx\n", src_base); + else { + printk("."); + if (!(k++ % 64)) + printk("\n"); + } + for (i = 0; i < comp_pages; i++) { + dest = crash_alloc_dest_page(); + if (crash_debug) { + printk("%lx ", dest); + if (!(j++ % 8)) + printk("\n"); + } + header->data_pages++; + if (!dest) { + printk("save_core: no dest pages\n"); + return; + } + if (!crash_is_kseg(dest) || !crash_is_kseg(cp)) { + printk("dest = 0x%lx, cp = 0x%lx, i = %d, comp_pages = 0x%lx\n", + dest, cp, i, comp_pages); + src = 0; + break; + } + memcpy((void *)dest, (void *)cp, PAGE_SIZE); + cp += PAGE_SIZE; + map = crash_update_map(map, src_base, dest, &header->map_pages); /* links a new map page, if necessary */ + if (!map) { + printk("save_core: no map\n"); + return; + } + } + header->compr_units++; + if (!src) + break; + if (crash_reset_stream(&stream)) + return; + } + + map = crash_update_map(map, 0, 0, &header->map_pages); + header->magic[0] = CRASH_MAGIC; + + if (crash_audit_maps()) { + header->magic[0] = 0; + return; + } + + printk("\nsave_core: src pages skipped = 0x%lx src pages saved = 0x%lx\n", + src_pages_skipped, src_pages_saved); + printk("save_core: data_pages = 0x%lx map_pages = 0x%lx\n", header->data_pages, + header->map_pages); + printk("save_core: completed, crash_dump_header = 0x%lx\n", crash_dump_header); +} + +/* helper functions private to this file */ +static int crash_reset_stream(z_stream * stream) +{ + int err; + + stream->zalloc = czalloc; + stream->zfree = czfree; + stream->opaque = (voidpf) 0; + stream->next_out = (Bytef *) crash_compr_buf; + stream->avail_out = (uInt) (CRASH_COMPR_BUF_PAGES * PAGE_SIZE); + stream->next_in = (Bytef *) crash_uncompr_buf; + stream->avail_in = (uInt) (CRASH_UNCOMPR_BUF_PAGES * PAGE_SIZE); + err = deflateReset(stream); + if (err != Z_OK) { + printk("crash_reset_stream: bad return %d from deflateReset\n", err); + return 1; + } + return 0; +} + +static u_long crash_alloc_dest_page(void) +{ + u_long addr; + + addr = crash_dest_free_list; + if (addr) { + crash_dest_free_list = *(u_long *) addr; + dest_pages_free--; + } else + printk("crash_alloc_dest_page: free list empty\n"); + return addr; +} + +static void crash_free_dest_page(u_long dest) +{ + if (!dest) { + printk("crash_free_dest_page: freeing addr 0\n"); + return; + } + dest_pages_free++; + dest = (u_long) trunc_page(dest); + *(u_long *) dest = crash_dest_free_list; + crash_dest_free_list = dest; +} + +/* + * Stolen from setup.c + */ +#define PFN_PHYS(x) ((x) << PAGE_SHIFT) + +static void init_dest_page_alloc(void) +{ + u_long va; + long i; + struct page *page; + struct mem_crash_map_hdr *header; + + header = (struct mem_crash_map_hdr *)crash_dump_header; + for (i = ((1 << 24) >> PAGE_SHIFT) + LOWER_MEM_FORWARD; + i < (max_low_pfn - UPPER_MEM_BACKUP); i++) { + va = (u_long) phys_to_virt(PFN_PHYS(i)); + if ((va >= header->boot_reserved_start) && (va < header->boot_reserved_end)) + continue; + page = mem_map + i; + if (PageLocked(page) || PageReserved(page)) + continue; + if (PageFree(page) || PageAnon(page) || PageShm(page) || page->buffers) + crash_free_dest_page(va); + } + if (crash_debug) + printk("init_dest_page_alloc: dest_pages_free = 0x%lx\n", dest_pages_free); +} + +static int crash_is_dir_page(struct page *page) { + struct inode *tmp_inode; + + if(page->mapping && page->mapping->host) { + tmp_inode = (struct inode *)page->mapping->host; + if((tmp_inode->i_sb->s_magic == EXT2_SUPER_MAGIC) && + (S_ISDIR(tmp_inode->i_mode))) + return 1; + } + + return 0; +} + +static u_long crash_get_source_page(void) +{ + struct page *page; + u_long va; + + while (crash_cur_pfn < max_low_pfn) { + page = mem_map + crash_cur_pfn; + if (!(PageFree(page) || PageAnon(page) || PageShm(page) || page->buffers)) + break; + src_pages_skipped++; + crash_cur_pfn++; + } + if (crash_cur_pfn == max_low_pfn) + return 0; + + va = (u_long) phys_to_virt(PFN_PHYS(crash_cur_pfn)); + src_pages_saved++; + crash_cur_pfn++; + return va; +} + +static u_long crash_update_map(u_long map, u_long src_base, u_long dest, u_long * pages) +{ + struct mem_crash_map_entry *m; + + + if (!map) { + (*pages)++; + return crash_alloc_dest_page(); + } + m = (struct mem_crash_map_entry *)map; + m->src_va = src_base; + m->dest_page_va = dest; + if (dest) + if (crash_chksum_page(dest, &m->check_sum)) + return 0; + + map += sizeof(struct mem_crash_map_entry); + + m = (struct mem_crash_map_entry *)map; + if (!src_base) { /* end of list */ + if (crash_chksum((u_long) m, &m->src_va)) + return 0; + } else if ((map + 3 * sizeof(struct mem_crash_map_entry)) > (u_long) round_page(map)) { + m->src_va = -1; + map = m->dest_page_va = crash_alloc_dest_page(); + if (crash_debug) + printk("\nm = 0x%lx m->src_va = 0x%lx m->dest_page_va = 0x%lx\n", + (u_long) trunc_page(m), m->src_va, m->dest_page_va); + m++; + if (crash_chksum((u_long) m, &m->src_va)) + return 0; + if (crash_debug) + printk("m = 0x%lx chksum = m->src_va = 0x%lx\n", (u_long) trunc_page(m), + m->src_va); + if (crash_audit_map_page((u_long) m)) + return 0; + (*pages)++; + } + return map; +} + +static int crash_chksum(u_long limit, u_long * sum_addr) +{ + u_long sum; + u_long *addr; + + if (!crash_is_kseg(limit)) { + printk("bad addr = 0x%lx to crash_chksum\n", limit); + return 1; + } + sum = 0; + addr = (u_long *) trunc_page(limit); + for (; (u_long) addr < limit; addr++) + sum += *addr; + *sum_addr = sum; + return 0; +} + +static int crash_chksum_page(u_long pg_addr, u_long * sum_addr) +{ + u_long sum, limit; + u_long *addr; + + if (!crash_is_kseg(pg_addr)) { + printk("bad addr = 0x%lx to crash_chksum_page\n", pg_addr); + return 1; + } + + sum = 0; + addr = (u_long *) trunc_page(pg_addr); + limit = (u_long) addr + PAGE_SIZE; + for (; (u_long) addr < limit; addr++) + sum += *addr; + *sum_addr = sum; + return 0; +} + +static int crash_audit_maps(void) +{ + u_long m, count; + u_long *link_addr; + struct mem_crash_map_hdr *header; + + header = (struct mem_crash_map_hdr *)crash_dump_header; + if (header->magic[0] != CRASH_MAGIC) + return 1; + + link_addr = &header->map; + m = header->map; + + count = 0; + for (;;) { + if (!crash_is_kseg(m)) { + printk("crash_audit_maps: bad link 0x%lx at 0x%lx\n", m, + (u_long) link_addr); + return 1; + } + if (crash_audit_map_page(m)) { + printk("audit failed while on map page %ld\n", count); + return 1; + } + if (!crash_link(m)) + break; + link_addr = crash_link(m); + m = *link_addr; + + count++; + } + return 0; +} + +static int crash_audit_map_page(u_long map) +{ + struct mem_crash_map_entry *m; + u_long sum; + + if (!map || !crash_is_kseg(map)) { + printk("crash_audit_map_page: bad map = 0x%lx\n", map); + return 1; + } + map = (u_long) trunc_page((u_long) map); + m = (struct mem_crash_map_entry *)map; + for (;;) { + if ((m->src_va == -1) || (m->src_va == 0)) { + m++; + if (crash_chksum((u_long) m, &sum)) + return 1; + if (m->src_va != sum) { + printk("crash_audit_map_page: checksum failure1\n"); + printk("m = 0x%lx, sum = 0x%lx, m->src_va = 0x%lx\n", + (u_long) m, (u_long) sum, (u_long) m->src_va); + crash_print_data_around((u_long) & m->src_va); + return 1; + } else { + return 0; + } + } else { + if (crash_chksum_page((u_long) m->dest_page_va, &sum) + || (m->check_sum != sum)) { + printk("crash_audit_map_page: checksum failure2\n"); + printk + ("dest_page_va = 0x%lx, &dest_page_va = 0x%lx, sum = 0x%lx, m->check_sum = 0x%lx\n", + (u_long) m->dest_page_va, (u_long) (&m->check_sum), + (u_long) sum, (u_long) m->check_sum); + crash_print_data_around((u_long) & m->check_sum); + return 1; + } + } + m++; + } +} + +static void crash_print_data_around(u_long p) +{ + u_long *a; + int i; + + if (!crash_is_kseg(p)) { + printk("crash_print_data_around: p = 0x%lx not kseg\n", p); + return; + } + a = (u_long *) p; + a -= 20; + for (i = 0; i < 40; i++) + printk("%lx\n", *a++); +} + +#ifdef CRASH_DEBUG +static void crash_print_map_page(u_long map) +{ + struct mem_crash_map_entry *m; + int j = 0; + u_long sum; + + map = (u_long) trunc_page((u_long) map); + m = (struct mem_crash_map_entry *)map; + for (;;) { + printk("%lx %lx %lx ", m->src_va, m->dest_page_va, m->check_sum); + if (!(j++ % 4)) + printk("\n"); + if ((m->src_va == -1) || (m->src_va == 0)) { + m++; + printk("%lx %lx ", m->src_va, m->dest_page_va); + if (crash_chksum((u_long) m, &sum)); + else + printk("\nchksum = 0x%lx\n", sum); + return; + } + m++; + } +} +#endif /* CRASH_DEBUG */ + +static void crash_wait_cpus(void) +{ + int i; + int msecs = 0; + + for (i = 0; i < smp_num_cpus; i++) { + if (i != smp_processor_id()) { + while (!panic_threads[i]) { + msecs++; + mdelay(1); + if (msecs > CRASH_CPU_TIMEOUT) { + /* if other cpus are still running + * we have to halt, otherwise we could + * risk using buffer cache pages which + * could subsequently get flushed to disk. + */ + printk("Unable to halt other CPUs, halting system.\n"); + crash_halt_or_reboot(0); + } + } + } + } + + crash_cleanup_smp_state(); +} + + +static void *czalloc(void *arg, unsigned int items, unsigned int size) +{ + u_long nbytes; + u_long addr; + + nbytes = (u_long) (items * size); + nbytes = (u_long) round_page(nbytes); + if ((crash_zalloc_cur + nbytes) > crash_zalloc_end) + return 0; + addr = crash_zalloc_cur; + crash_zalloc_cur += nbytes; + return ((void *)addr); +} + +static void czfree(void *arg, void *ptr) +{ + printk("zfree: ptr = 0x%lx\n", (u_long) ptr); +} + +static boolean_t crash_is_kseg(u_long addr) +{ + u_long phys; + + phys = virt_to_phys((void *)addr); + if (phys < PFN_PHYS(max_low_pfn)) + return TRUE; + else + return FALSE; +} + +static u_long *crash_link(u_long p) +{ + struct mem_crash_map_entry *m; + + p = (u_long) trunc_page(p); + m = (struct mem_crash_map_entry *)p; + for (; m->src_va; m++) + if (m->src_va == -1) + return &m->dest_page_va; + + return 0; +} + +/* Call this after data written to disk. */ +static int crash_free_crashmem(void) +{ + struct mem_crash_map_hdr *header; + struct mem_crash_map_entry *m, *last_m; + + if (crash_debug) + printk("crash_free_crashmem: \n"); + + header = (struct mem_crash_map_hdr *)crash_dump_header; + if (crash_audit_maps()) { + header->magic[0] = 0; + return 1; + } + m = (struct mem_crash_map_entry *)header->map; + again: + for (; m->src_va; m++) { + if (m->src_va == -1) { + last_m = m; + m = (struct mem_crash_map_entry *)m->dest_page_va; + crash_free_page((unsigned long)last_m); + goto again; + } + crash_free_page(m->dest_page_va); + } + if (crash_debug) + printk("crash_free_crashmem: 0x%lx freed\n", + (header->data_pages + header->map_pages) * PAGE_SIZE); + header->magic[0] = 0; + return 0; +} + +static void crash_free_page(u_long addr) +{ + struct page *page; + + page = virt_to_page(addr); + ClearPageReserved(page); + set_page_count(page, 1); + __free_page(page); +} + +static int get_dump_helper(u_long kva, u_long buf) +{ + struct page *page; + struct mem_crash_map_hdr *header; + + header = (struct mem_crash_map_hdr *)crash_dump_header; + if (header->magic[0] != CRASH_MAGIC) + return 1; + + if (!kva) { + if (crash_audit_maps()) { + printk("get_dump_helper: audit failure\n"); + header->magic[0] = 0; + return 1; + } + page = virt_to_page((u_long) crash_dump_header); + if (!PageReserved(page)) { + printk("not reserved: crash_dump_header = 0x%lx\n", crash_dump_header); + return 1; + } + if (copy_to_user((char *)buf, (char *)crash_dump_header, + sizeof(struct mem_crash_map_hdr))) { + printk("get_dump_helper: copy_to_user failed1\n"); + return 1; + } + } else { + page = virt_to_page(kva); + if (!PageReserved(page)) { + printk("not reserved: kva = 0x%lx\n", kva); + return 1; + } + if (copy_to_user((char *)buf, (char *)trunc_page(kva), PAGE_SIZE)) { + printk("get_dump_helper: copy_to_user failed2\n"); + return 1; + } + } + return 0; +} + +static void free_dump_helper(void) +{ + struct mem_crash_map_hdr *header; + + header = (struct mem_crash_map_hdr *)crash_dump_header; + if (header->magic[0] != CRASH_MAGIC) + return; + if (crash_debug) + printk("free_dump_helper\n"); + crash_free_crashmem(); +} + +static int crashdev_open(struct inode *inode, struct file *file) +{ + /* always return success -- nothing to do here */ + return 0; +} + +/* character device implementation */ +static struct file_operations crashdev_fops = { + ioctl:crashdev_ioctl, + open:crashdev_open, +}; + +static struct miscdevice crash_miscdev = { + 190, "crash", &crashdev_fops +}; + +int crash_init_chrdev(void) +{ + int result; + + result = misc_register(&crash_miscdev); + + if (result < 0) + printk(KERN_WARNING "crash: can't register crash device (c 10 190)\n"); + + return result; +} + +/* call the original syscalls, just to get things going */ +static int crashdev_ioctl(struct inode *inode, struct file *file, + unsigned int cmd, unsigned long arg) +{ + int retval = 0; + + switch (cmd) { + case CRASH_IOCFREEDUMP: + free_dump_helper(); + break; + + case CRASH_IOCGETDUMP: + if (crash_debug) { + printk("crashdev_ioctl: get dump\n"); + printk("vals: %08lx %08lx\n", + ((struct ioctl_getdump *)arg)->kva, + ((struct ioctl_getdump *)arg)->buf); + } + + retval = get_dump_helper((u_long) ((struct ioctl_getdump *)arg)->kva, + (u_long) ((struct ioctl_getdump *)arg)->buf); + break; + +#ifdef CONFIG_BOOTIMG + case CRASH_IOCBOOTIMG: + if (crash_debug) + printk("crashdev_ioctl: bootimg\n"); + + retval = sys_bootimg((struct boot_image *)arg); + break; +#endif + + case CRASH_IOCVERSION: + if (crash_debug) + printk("crashdev_ioctl: version\n"); + retval = CRASH_K_MINOR | (CRASH_K_MAJOR << 16); + break; + + default: + return -EINVAL; + } + + return retval; +} diff -urN linux-2.4.7/kernel/panic.c linux-2.4.7-mcore/kernel/panic.c --- linux-2.4.7/kernel/panic.c Mon Oct 16 15:58:51 2000 +++ linux-2.4.7-mcore/kernel/panic.c Fri Jul 27 14:58:14 2001 @@ -17,6 +17,10 @@ #include #include +#ifdef CONFIG_MCL_COREDUMP +#include +#endif + asmlinkage void sys_sync(void); /* it's really int */ extern void unblank_console(void); @@ -51,18 +55,41 @@ unsigned long caller = (unsigned long) __builtin_return_address(0); #endif +#ifdef CONFIG_MCL_COREDUMP + crash_save_regs(); +#endif + va_start(args, fmt); vsprintf(buf, fmt, args); va_end(args); printk(KERN_EMERG "Kernel panic: %s\n",buf); + +#ifdef CONFIG_MCL_COREDUMP + if(!panicmsg) { + panicmsg = buf; + panic_processor = smp_processor_id(); + printk("panic processor = %d\n", panic_processor); + mb(); + } +#endif + if (in_interrupt()) printk(KERN_EMERG "In interrupt handler - not syncing\n"); else if (!current->pid) printk(KERN_EMERG "In idle task - not syncing\n"); - else +#ifdef CONFIG_MCL_COREDUMP + else if(crash_perform_sync) +#else + else +#endif sys_sync(); unblank_console(); + +#ifdef CONFIG_MCL_COREDUMP + smp_call_function((void*)smp_crash_funnel_cpu,0,0,0); + crash_save_current_state(current); +#endif #ifdef CONFIG_SMP smp_send_stop(); diff -urN linux-2.4.7/kernel/sysctl.c linux-2.4.7-mcore/kernel/sysctl.c --- linux-2.4.7/kernel/sysctl.c Thu Apr 12 15:20:31 2001 +++ linux-2.4.7-mcore/kernel/sysctl.c Fri Jul 27 14:58:14 2001 @@ -37,6 +37,10 @@ #include #endif +#ifdef CONFIG_MCL_COREDUMP +#include +#endif + #if defined(CONFIG_SYSCTL) /* External variables not in a header file. */ @@ -232,6 +236,10 @@ {KERN_SYSRQ, "sysrq", &sysrq_enabled, sizeof (int), 0644, NULL, &proc_dointvec}, #endif +#ifdef CONFIG_MCL_COREDUMP + {KERN_PANIC_ON_OOPS, "panic_on_oops", &panic_on_oops, sizeof(int), + 0644, NULL, &proc_dointvec}, +#endif {KERN_MAX_THREADS, "threads-max", &max_threads, sizeof(int), 0644, NULL, &proc_dointvec}, {KERN_RANDOM, "random", NULL, 0, 0555, random_table}, diff -urN linux-2.4.7/kernel/zlib.c linux-2.4.7-mcore/kernel/zlib.c --- linux-2.4.7/kernel/zlib.c Wed Dec 31 19:00:00 1969 +++ linux-2.4.7-mcore/kernel/zlib.c Fri Jul 27 14:58:15 2001 @@ -0,0 +1,5371 @@ +/* + * This file is derived from various .h and .c files from the zlib-1.0.4 + * distribution by Jean-loup Gailly and Mark Adler, with some additions + * by Paul Mackerras to aid in implementing Deflate compression and + * decompression for PPP packets. See zlib.h for conditions of + * distribution and use. + * + * Changes that have been made include: + * - added Z_PACKET_FLUSH (see zlib.h for details) + * - added inflateIncomp and deflateOutputPending + * - allow strm->next_out to be NULL, meaning discard the output + * + * $Id: zlib.c,v 1.3 1997/12/23 10:47:42 paulus Exp $ + */ + +/* + * ==FILEVERSION 971210== + * + * This marker is used by the Linux installation script to determine + * whether an up-to-date version of this file is already installed. + */ + +#define NO_DUMMY_DECL +#define NO_ZCFUNCS +#define MY_ZCALLOC + +#if defined(__FreeBSD__) && (defined(KERNEL) || defined(_KERNEL)) +#define inflate inflate_ppp /* FreeBSD already has an inflate :-( */ +#endif + + +/* +++ zutil.h */ +/* zutil.h -- internal interface and configuration of the compression library + * Copyright (C) 1995-1996 Jean-loup Gailly. + * For conditions of distribution and use, see copyright notice in zlib.h + */ + +/* WARNING: this file should *not* be used by applications. It is + part of the implementation of the compression library and is + subject to change. Applications should only use zlib.h. + */ + +/* From: zutil.h,v 1.16 1996/07/24 13:41:13 me Exp $ */ + +#ifndef _Z_UTIL_H +#define _Z_UTIL_H + +#include "zlib.h" + +#if defined(KERNEL) || defined(_KERNEL) +/* Assume this is a *BSD or SVR4 kernel */ +#include +#include +#include +# define HAVE_MEMCPY +# define memcpy(d, s, n) bcopy((s), (d), (n)) +# define memset(d, v, n) bzero((d), (n)) +# define memcmp bcmp + +#else +#if defined(__KERNEL__) +/* Assume this is a Linux kernel */ +#include +#define HAVE_MEMCPY + +#else /* not kernel */ + +#if defined(MSDOS)||defined(VMS)||defined(CRAY)||defined(WIN32)||defined(RISCOS) +# include +# include +#else + extern int errno; +#endif +#ifdef STDC +# include +# include +#endif +#endif /* __KERNEL__ */ +#endif /* _KERNEL || KERNEL */ + +#ifndef local +# define local static +#endif +/* compile with -Dlocal if your debugger can't find static symbols */ + +typedef unsigned char uch; +typedef uch FAR uchf; +typedef unsigned short ush; +typedef ush FAR ushf; +typedef unsigned long ulg; + +extern const char *z_errmsg[10]; /* indexed by 2-zlib_error */ +/* (size given to avoid silly warnings with Visual C++) */ + +#define ERR_MSG(err) z_errmsg[Z_NEED_DICT-(err)] + +#define ERR_RETURN(strm,err) \ + return (strm->msg = (char*)ERR_MSG(err), (err)) +/* To be used only when the state is known to be valid */ + + /* common constants */ + +#ifndef DEF_WBITS +# define DEF_WBITS MAX_WBITS +#endif +/* default windowBits for decompression. MAX_WBITS is for compression only */ + +#if MAX_MEM_LEVEL >= 8 +# define DEF_MEM_LEVEL 8 +#else +# define DEF_MEM_LEVEL MAX_MEM_LEVEL +#endif +/* default memLevel */ + +#define STORED_BLOCK 0 +#define STATIC_TREES 1 +#define DYN_TREES 2 +/* The three kinds of block type */ + +#define MIN_MATCH 3 +#define MAX_MATCH 258 +/* The minimum and maximum match lengths */ + +#define PRESET_DICT 0x20 /* preset dictionary flag in zlib header */ + + /* target dependencies */ + +#ifdef MSDOS +# define OS_CODE 0x00 +# ifdef __TURBOC__ +# include +# else /* MSC or DJGPP */ +# include +# endif +#endif + +#ifdef OS2 +# define OS_CODE 0x06 +#endif + +#ifdef WIN32 /* Window 95 & Windows NT */ +# define OS_CODE 0x0b +#endif + +#if defined(VAXC) || defined(VMS) +# define OS_CODE 0x02 +# define FOPEN(name, mode) \ + fopen((name), (mode), "mbc=60", "ctx=stm", "rfm=fix", "mrs=512") +#endif + +#ifdef AMIGA +# define OS_CODE 0x01 +#endif + +#if defined(ATARI) || defined(atarist) +# define OS_CODE 0x05 +#endif + +#ifdef MACOS +# define OS_CODE 0x07 +#endif + +#ifdef __50SERIES /* Prime/PRIMOS */ +# define OS_CODE 0x0F +#endif + +#ifdef TOPS20 +# define OS_CODE 0x0a +#endif + +#if defined(_BEOS_) || defined(RISCOS) +# define fdopen(fd,mode) NULL /* No fdopen() */ +#endif + + /* Common defaults */ + +#ifndef OS_CODE +# define OS_CODE 0x03 /* assume Unix */ +#endif + +#ifndef FOPEN +# define FOPEN(name, mode) fopen((name), (mode)) +#endif + + /* functions */ + +#ifdef HAVE_STRERROR + extern char *strerror OF((int)); +# define zstrerror(errnum) strerror(errnum) +#else +# define zstrerror(errnum) "" +#endif + +#if defined(pyr) +# define NO_MEMCPY +#endif +#if (defined(M_I86SM) || defined(M_I86MM)) && !defined(_MSC_VER) + /* Use our own functions for small and medium model with MSC <= 5.0. + * You may have to use the same strategy for Borland C (untested). + */ +# define NO_MEMCPY +#endif +#if defined(STDC) && !defined(HAVE_MEMCPY) && !defined(NO_MEMCPY) +# define HAVE_MEMCPY +#endif +#ifdef HAVE_MEMCPY +# ifdef SMALL_MEDIUM /* MSDOS small or medium model */ +# define zmemcpy _fmemcpy +# define zmemcmp _fmemcmp +# define zmemzero(dest, len) _fmemset(dest, 0, len) +# else +# define zmemcpy memcpy +# define zmemcmp memcmp +# define zmemzero(dest, len) memset(dest, 0, len) +# endif +#else + extern void zmemcpy OF((Bytef* dest, Bytef* source, uInt len)); + extern int zmemcmp OF((Bytef* s1, Bytef* s2, uInt len)); + extern void zmemzero OF((Bytef* dest, uInt len)); +#endif + +/* Diagnostic functions */ +#ifdef DEBUG_ZLIB +# include +# ifndef verbose +# define verbose 0 +# endif + extern void z_error OF((char *m)); +# define Assert(cond,msg) {if(!(cond)) z_error(msg);} +# define Trace(x) fprintf x +# define Tracev(x) {if (verbose) fprintf x ;} +# define Tracevv(x) {if (verbose>1) fprintf x ;} +# define Tracec(c,x) {if (verbose && (c)) fprintf x ;} +# define Tracecv(c,x) {if (verbose>1 && (c)) fprintf x ;} +#else +# define Assert(cond,msg) +# define Trace(x) +# define Tracev(x) +# define Tracevv(x) +# define Tracec(c,x) +# define Tracecv(c,x) +#endif + + +typedef uLong (*check_func) OF((uLong check, const Bytef *buf, uInt len)); + +voidpf zcalloc OF((voidpf opaque, unsigned items, unsigned size)); +void zcfree OF((voidpf opaque, voidpf ptr)); + +#define ZALLOC(strm, items, size) \ + (*((strm)->zalloc))((strm)->opaque, (items), (size)) +#define ZFREE(strm, addr) (*((strm)->zfree))((strm)->opaque, (voidpf)(addr)) +#define TRY_FREE(s, p) {if (p) ZFREE(s, p);} + +#endif /* _Z_UTIL_H */ +/* --- zutil.h */ + +/* +++ deflate.h */ +/* deflate.h -- internal compression state + * Copyright (C) 1995-1996 Jean-loup Gailly + * For conditions of distribution and use, see copyright notice in zlib.h + */ + +/* WARNING: this file should *not* be used by applications. It is + part of the implementation of the compression library and is + subject to change. Applications should only use zlib.h. + */ + +/* From: deflate.h,v 1.10 1996/07/02 12:41:00 me Exp $ */ + +#ifndef _DEFLATE_H +#define _DEFLATE_H + +/* #include "zutil.h" */ + +/* =========================================================================== + * Internal compression state. + */ + +#define LENGTH_CODES 29 +/* number of length codes, not counting the special END_BLOCK code */ + +#define LITERALS 256 +/* number of literal bytes 0..255 */ + +#define L_CODES (LITERALS+1+LENGTH_CODES) +/* number of Literal or Length codes, including the END_BLOCK code */ + +#define D_CODES 30 +/* number of distance codes */ + +#define BL_CODES 19 +/* number of codes used to transfer the bit lengths */ + +#define HEAP_SIZE (2*L_CODES+1) +/* maximum heap size */ + +#define MAX_BITS 15 +/* All codes must not exceed MAX_BITS bits */ + +#define INIT_STATE 42 +#define BUSY_STATE 113 +#define FINISH_STATE 666 +/* Stream status */ + + +/* Data structure describing a single value and its code string. */ +typedef struct ct_data_s { + union { + ush freq; /* frequency count */ + ush code; /* bit string */ + } fc; + union { + ush dad; /* father node in Huffman tree */ + ush len; /* length of bit string */ + } dl; +} FAR ct_data; + +#define Freq fc.freq +#define Code fc.code +#define Dad dl.dad +#define Len dl.len + +typedef struct static_tree_desc_s static_tree_desc; + +typedef struct tree_desc_s { + ct_data *dyn_tree; /* the dynamic tree */ + int max_code; /* largest code with non zero frequency */ + static_tree_desc *stat_desc; /* the corresponding static tree */ +} FAR tree_desc; + +typedef ush Pos; +typedef Pos FAR Posf; +typedef unsigned IPos; + +/* A Pos is an index in the character window. We use short instead of int to + * save space in the various tables. IPos is used only for parameter passing. + */ + +typedef struct deflate_state { + z_streamp strm; /* pointer back to this zlib stream */ + int status; /* as the name implies */ + Bytef *pending_buf; /* output still pending */ + ulg pending_buf_size; /* size of pending_buf */ + Bytef *pending_out; /* next pending byte to output to the stream */ + int pending; /* nb of bytes in the pending buffer */ + int noheader; /* suppress zlib header and adler32 */ + Byte data_type; /* UNKNOWN, BINARY or ASCII */ + Byte method; /* STORED (for zip only) or DEFLATED */ + int last_flush; /* value of flush param for previous deflate call */ + + /* used by deflate.c: */ + + uInt w_size; /* LZ77 window size (32K by default) */ + uInt w_bits; /* log2(w_size) (8..16) */ + uInt w_mask; /* w_size - 1 */ + + Bytef *window; + /* Sliding window. Input bytes are read into the second half of the window, + * and move to the first half later to keep a dictionary of at least wSize + * bytes. With this organization, matches are limited to a distance of + * wSize-MAX_MATCH bytes, but this ensures that IO is always + * performed with a length multiple of the block size. Also, it limits + * the window size to 64K, which is quite useful on MSDOS. + * To do: use the user input buffer as sliding window. + */ + + ulg window_size; + /* Actual size of window: 2*wSize, except when the user input buffer + * is directly used as sliding window. + */ + + Posf *prev; + /* Link to older string with same hash index. To limit the size of this + * array to 64K, this link is maintained only for the last 32K strings. + * An index in this array is thus a window index modulo 32K. + */ + + Posf *head; /* Heads of the hash chains or NIL. */ + + uInt ins_h; /* hash index of string to be inserted */ + uInt hash_size; /* number of elements in hash table */ + uInt hash_bits; /* log2(hash_size) */ + uInt hash_mask; /* hash_size-1 */ + + uInt hash_shift; + /* Number of bits by which ins_h must be shifted at each input + * step. It must be such that after MIN_MATCH steps, the oldest + * byte no longer takes part in the hash key, that is: + * hash_shift * MIN_MATCH >= hash_bits + */ + + long block_start; + /* Window position at the beginning of the current output block. Gets + * negative when the window is moved backwards. + */ + + uInt match_length; /* length of best match */ + IPos prev_match; /* previous match */ + int match_available; /* set if previous match exists */ + uInt strstart; /* start of string to insert */ + uInt match_start; /* start of matching string */ + uInt lookahead; /* number of valid bytes ahead in window */ + + uInt prev_length; + /* Length of the best match at previous step. Matches not greater than this + * are discarded. This is used in the lazy match evaluation. + */ + + uInt max_chain_length; + /* To speed up deflation, hash chains are never searched beyond this + * length. A higher limit improves compression ratio but degrades the + * speed. + */ + + uInt max_lazy_match; + /* Attempt to find a better match only when the current match is strictly + * smaller than this value. This mechanism is used only for compression + * levels >= 4. + */ +# define max_insert_length max_lazy_match + /* Insert new strings in the hash table only if the match length is not + * greater than this length. This saves time but degrades compression. + * max_insert_length is used only for compression levels <= 3. + */ + + int level; /* compression level (1..9) */ + int strategy; /* favor or force Huffman coding*/ + + uInt good_match; + /* Use a faster search when the previous match is longer than this */ + + int nice_match; /* Stop searching when current match exceeds this */ + + /* used by trees.c: */ + /* Didn't use ct_data typedef below to supress compiler warning */ + struct ct_data_s dyn_ltree[HEAP_SIZE]; /* literal and length tree */ + struct ct_data_s dyn_dtree[2*D_CODES+1]; /* distance tree */ + struct ct_data_s bl_tree[2*BL_CODES+1]; /* Huffman tree for bit lengths */ + + struct tree_desc_s l_desc; /* desc. for literal tree */ + struct tree_desc_s d_desc; /* desc. for distance tree */ + struct tree_desc_s bl_desc; /* desc. for bit length tree */ + + ush bl_count[MAX_BITS+1]; + /* number of codes at each bit length for an optimal tree */ + + int heap[2*L_CODES+1]; /* heap used to build the Huffman trees */ + int heap_len; /* number of elements in the heap */ + int heap_max; /* element of largest frequency */ + /* The sons of heap[n] are heap[2*n] and heap[2*n+1]. heap[0] is not used. + * The same heap array is used to build all trees. + */ + + uch depth[2*L_CODES+1]; + /* Depth of each subtree used as tie breaker for trees of equal frequency + */ + + uchf *l_buf; /* buffer for literals or lengths */ + + uInt lit_bufsize; + /* Size of match buffer for literals/lengths. There are 4 reasons for + * limiting lit_bufsize to 64K: + * - frequencies can be kept in 16 bit counters + * - if compression is not successful for the first block, all input + * data is still in the window so we can still emit a stored block even + * when input comes from standard input. (This can also be done for + * all blocks if lit_bufsize is not greater than 32K.) + * - if compression is not successful for a file smaller than 64K, we can + * even emit a stored file instead of a stored block (saving 5 bytes). + * This is applicable only for zip (not gzip or zlib). + * - creating new Huffman trees less frequently may not provide fast + * adaptation to changes in the input data statistics. (Take for + * example a binary file with poorly compressible code followed by + * a highly compressible string table.) Smaller buffer sizes give + * fast adaptation but have of course the overhead of transmitting + * trees more frequently. + * - I can't count above 4 + */ + + uInt last_lit; /* running index in l_buf */ + + ushf *d_buf; + /* Buffer for distances. To simplify the code, d_buf and l_buf have + * the same number of elements. To use different lengths, an extra flag + * array would be necessary. + */ + + ulg opt_len; /* bit length of current block with optimal trees */ + ulg static_len; /* bit length of current block with static trees */ + ulg compressed_len; /* total bit length of compressed file */ + uInt matches; /* number of string matches in current block */ + int last_eob_len; /* bit length of EOB code for last block */ + +#ifdef DEBUG_ZLIB + ulg bits_sent; /* bit length of the compressed data */ +#endif + + ush bi_buf; + /* Output buffer. bits are inserted starting at the bottom (least + * significant bits). + */ + int bi_valid; + /* Number of valid bits in bi_buf. All bits above the last valid bit + * are always zero. + */ + +} FAR deflate_state; + +/* Output a byte on the stream. + * IN assertion: there is enough room in pending_buf. + */ +#define put_byte(s, c) {s->pending_buf[s->pending++] = (c);} + + +#define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1) +/* Minimum amount of lookahead, except at the end of the input file. + * See deflate.c for comments about the MIN_MATCH+1. + */ + +#define MAX_DIST(s) ((s)->w_size-MIN_LOOKAHEAD) +/* In order to simplify the code, particularly on 16 bit machines, match + * distances are limited to MAX_DIST instead of WSIZE. + */ + + /* in trees.c */ +void _tr_init OF((deflate_state *s)); +int _tr_tally OF((deflate_state *s, unsigned dist, unsigned lc)); +ulg _tr_flush_block OF((deflate_state *s, charf *buf, ulg stored_len, + int eof)); +void _tr_align OF((deflate_state *s)); +void _tr_stored_block OF((deflate_state *s, charf *buf, ulg stored_len, + int eof)); +void _tr_stored_type_only OF((deflate_state *)); + +#endif +/* --- deflate.h */ + +/* +++ deflate.c */ +/* deflate.c -- compress data using the deflation algorithm + * Copyright (C) 1995-1996 Jean-loup Gailly. + * For conditions of distribution and use, see copyright notice in zlib.h + */ + +/* + * ALGORITHM + * + * The "deflation" process depends on being able to identify portions + * of the input text which are identical to earlier input (within a + * sliding window trailing behind the input currently being processed). + * + * The most straightforward technique turns out to be the fastest for + * most input files: try all possible matches and select the longest. + * The key feature of this algorithm is that insertions into the string + * dictionary are very simple and thus fast, and deletions are avoided + * completely. Insertions are performed at each input character, whereas + * string matches are performed only when the previous match ends. So it + * is preferable to spend more time in matches to allow very fast string + * insertions and avoid deletions. The matching algorithm for small + * strings is inspired from that of Rabin & Karp. A brute force approach + * is used to find longer strings when a small match has been found. + * A similar algorithm is used in comic (by Jan-Mark Wams) and freeze + * (by Leonid Broukhis). + * A previous version of this file used a more sophisticated algorithm + * (by Fiala and Greene) which is guaranteed to run in linear amortized + * time, but has a larger average cost, uses more memory and is patented. + * However the F&G algorithm may be faster for some highly redundant + * files if the parameter max_chain_length (described below) is too large. + * + * ACKNOWLEDGEMENTS + * + * The idea of lazy evaluation of matches is due to Jan-Mark Wams, and + * I found it in 'freeze' written by Leonid Broukhis. + * Thanks to many people for bug reports and testing. + * + * REFERENCES + * + * Deutsch, L.P.,"DEFLATE Compressed Data Format Specification". + * Available in ftp://ds.internic.net/rfc/rfc1951.txt + * + * A description of the Rabin and Karp algorithm is given in the book + * "Algorithms" by R. Sedgewick, Addison-Wesley, p252. + * + * Fiala,E.R., and Greene,D.H. + * Data Compression with Finite Windows, Comm.ACM, 32,4 (1989) 490-595 + * + */ + +/* From: deflate.c,v 1.15 1996/07/24 13:40:58 me Exp $ */ + +/* #include "deflate.h" */ + +char deflate_copyright[] = " deflate 1.0.4 Copyright 1995-1996 Jean-loup Gailly "; +/* + If you use the zlib library in a product, an acknowledgment is welcome + in the documentation of your product. If for some reason you cannot + include such an acknowledgment, I would appreciate that you keep this + copyright string in the executable of your product. + */ + +/* =========================================================================== + * Function prototypes. + */ +typedef enum { + need_more, /* block not completed, need more input or more output */ + block_done, /* block flush performed */ + finish_started, /* finish started, need only more output at next deflate */ + finish_done /* finish done, accept no more input or output */ +} block_state; + +typedef block_state (*compress_func) OF((deflate_state *s, int flush)); +/* Compression function. Returns the block state after the call. */ + +local void fill_window OF((deflate_state *s)); +local block_state deflate_stored OF((deflate_state *s, int flush)); +local block_state deflate_fast OF((deflate_state *s, int flush)); +local block_state deflate_slow OF((deflate_state *s, int flush)); +local void lm_init OF((deflate_state *s)); +local void putShortMSB OF((deflate_state *s, uInt b)); +local void flush_pending OF((z_streamp strm)); +local int read_buf OF((z_streamp strm, charf *buf, unsigned size)); +#ifdef ASMV + void match_init OF((void)); /* asm code initialization */ + uInt longest_match OF((deflate_state *s, IPos cur_match)); +#else +local uInt longest_match OF((deflate_state *s, IPos cur_match)); +#endif + +#ifdef DEBUG_ZLIB +local void check_match OF((deflate_state *s, IPos start, IPos match, + int length)); +#endif + +/* =========================================================================== + * Local data + */ + +#define NIL 0 +/* Tail of hash chains */ + +#ifndef TOO_FAR +# define TOO_FAR 4096 +#endif +/* Matches of length 3 are discarded if their distance exceeds TOO_FAR */ + +#define MIN_LOOKAHEAD (MAX_MATCH+MIN_MATCH+1) +/* Minimum amount of lookahead, except at the end of the input file. + * See deflate.c for comments about the MIN_MATCH+1. + */ + +/* Values for max_lazy_match, good_match and max_chain_length, depending on + * the desired pack level (0..9). The values given below have been tuned to + * exclude worst case performance for pathological files. Better values may be + * found for specific files. + */ +typedef struct config_s { + ush good_length; /* reduce lazy search above this match length */ + ush max_lazy; /* do not perform lazy search above this match length */ + ush nice_length; /* quit search above this match length */ + ush max_chain; + compress_func func; +} config; + +local config configuration_table[10] = { +/* good lazy nice chain */ +/* 0 */ {0, 0, 0, 0, deflate_stored}, /* store only */ +/* 1 */ {4, 4, 8, 4, deflate_fast}, /* maximum speed, no lazy matches */ +/* 2 */ {4, 5, 16, 8, deflate_fast}, +/* 3 */ {4, 6, 32, 32, deflate_fast}, + +/* 4 */ {4, 4, 16, 16, deflate_slow}, /* lazy matches */ +/* 5 */ {8, 16, 32, 32, deflate_slow}, +/* 6 */ {8, 16, 128, 128, deflate_slow}, +/* 7 */ {8, 32, 128, 256, deflate_slow}, +/* 8 */ {32, 128, 258, 1024, deflate_slow}, +/* 9 */ {32, 258, 258, 4096, deflate_slow}}; /* maximum compression */ + +/* Note: the deflate() code requires max_lazy >= MIN_MATCH and max_chain >= 4 + * For deflate_fast() (levels <= 3) good is ignored and lazy has a different + * meaning. + */ + +#define EQUAL 0 +/* result of memcmp for equal strings */ + +#ifndef NO_DUMMY_DECL +struct static_tree_desc_s {int dummy;}; /* for buggy compilers */ +#endif + +/* =========================================================================== + * Update a hash value with the given input byte + * IN assertion: all calls to to UPDATE_HASH are made with consecutive + * input characters, so that a running hash key can be computed from the + * previous key instead of complete recalculation each time. + */ +#define UPDATE_HASH(s,h,c) (h = (((h)<hash_shift) ^ (c)) & s->hash_mask) + + +/* =========================================================================== + * Insert string str in the dictionary and set match_head to the previous head + * of the hash chain (the most recent string with same hash key). Return + * the previous length of the hash chain. + * IN assertion: all calls to to INSERT_STRING are made with consecutive + * input characters and the first MIN_MATCH bytes of str are valid + * (except for the last MIN_MATCH-1 bytes of the input file). + */ +#define INSERT_STRING(s, str, match_head) \ + (UPDATE_HASH(s, s->ins_h, s->window[(str) + (MIN_MATCH-1)]), \ + s->prev[(str) & s->w_mask] = match_head = s->head[s->ins_h], \ + s->head[s->ins_h] = (Pos)(str)) + +/* =========================================================================== + * Initialize the hash table (avoiding 64K overflow for 16 bit systems). + * prev[] will be initialized on the fly. + */ +#define CLEAR_HASH(s) \ + s->head[s->hash_size-1] = NIL; \ + zmemzero((charf *)s->head, (unsigned)(s->hash_size-1)*sizeof(*s->head)); + +/* ========================================================================= */ +int deflateInit_(strm, level, version, stream_size) + z_streamp strm; + int level; + const char *version; + int stream_size; +{ + return deflateInit2_(strm, level, Z_DEFLATED, MAX_WBITS, DEF_MEM_LEVEL, + Z_DEFAULT_STRATEGY, version, stream_size); + /* To do: ignore strm->next_in if we use it as window */ +} + +/* ========================================================================= */ +int deflateInit2_(strm, level, method, windowBits, memLevel, strategy, + version, stream_size) + z_streamp strm; + int level; + int method; + int windowBits; + int memLevel; + int strategy; + const char *version; + int stream_size; +{ + deflate_state *s; + int noheader = 0; + static char* my_version = ZLIB_VERSION; + + ushf *overlay; + /* We overlay pending_buf and d_buf+l_buf. This works since the average + * output size for (length,distance) codes is <= 24 bits. + */ + + if (version == Z_NULL || version[0] != my_version[0] || + stream_size != sizeof(z_stream)) { + return Z_VERSION_ERROR; + } + if (strm == Z_NULL) return Z_STREAM_ERROR; + + strm->msg = Z_NULL; +#ifndef NO_ZCFUNCS + if (strm->zalloc == Z_NULL) { + strm->zalloc = zcalloc; + strm->opaque = (voidpf)0; + } + if (strm->zfree == Z_NULL) strm->zfree = zcfree; +#endif + + if (level == Z_DEFAULT_COMPRESSION) level = 6; + + if (windowBits < 0) { /* undocumented feature: suppress zlib header */ + noheader = 1; + windowBits = -windowBits; + } + if (memLevel < 1 || memLevel > MAX_MEM_LEVEL || method != Z_DEFLATED || + windowBits < 8 || windowBits > 15 || level < 0 || level > 9 || + strategy < 0 || strategy > Z_HUFFMAN_ONLY) { + return Z_STREAM_ERROR; + } + s = (deflate_state *) ZALLOC(strm, 1, sizeof(deflate_state)); + if (s == Z_NULL) return Z_MEM_ERROR; + strm->state = (struct internal_state FAR *)s; + s->strm = strm; + + s->noheader = noheader; + s->w_bits = windowBits; + s->w_size = 1 << s->w_bits; + s->w_mask = s->w_size - 1; + + s->hash_bits = memLevel + 7; + s->hash_size = 1 << s->hash_bits; + s->hash_mask = s->hash_size - 1; + s->hash_shift = ((s->hash_bits+MIN_MATCH-1)/MIN_MATCH); + + s->window = (Bytef *) ZALLOC(strm, s->w_size, 2*sizeof(Byte)); + s->prev = (Posf *) ZALLOC(strm, s->w_size, sizeof(Pos)); + s->head = (Posf *) ZALLOC(strm, s->hash_size, sizeof(Pos)); + + s->lit_bufsize = 1 << (memLevel + 6); /* 16K elements by default */ + + overlay = (ushf *) ZALLOC(strm, s->lit_bufsize, sizeof(ush)+2); + s->pending_buf = (uchf *) overlay; + s->pending_buf_size = (ulg)s->lit_bufsize * (sizeof(ush)+2L); + + if (s->window == Z_NULL || s->prev == Z_NULL || s->head == Z_NULL || + s->pending_buf == Z_NULL) { + strm->msg = (char*)ERR_MSG(Z_MEM_ERROR); + deflateEnd (strm); + return Z_MEM_ERROR; + } + s->d_buf = overlay + s->lit_bufsize/sizeof(ush); + s->l_buf = s->pending_buf + (1+sizeof(ush))*s->lit_bufsize; + + s->level = level; + s->strategy = strategy; + s->method = (Byte)method; + + return deflateReset(strm); +} + +/* ========================================================================= */ +int deflateSetDictionary (strm, dictionary, dictLength) + z_streamp strm; + const Bytef *dictionary; + uInt dictLength; +{ + deflate_state *s; + uInt length = dictLength; + uInt n; + IPos hash_head = 0; + + if (strm == Z_NULL || strm->state == Z_NULL || dictionary == Z_NULL) + return Z_STREAM_ERROR; + + s = (deflate_state *) strm->state; + if (s->status != INIT_STATE) return Z_STREAM_ERROR; + + strm->adler = adler32(strm->adler, dictionary, dictLength); + + if (length < MIN_MATCH) return Z_OK; + if (length > MAX_DIST(s)) { + length = MAX_DIST(s); +#ifndef USE_DICT_HEAD + dictionary += dictLength - length; /* use the tail of the dictionary */ +#endif + } + zmemcpy((charf *)s->window, dictionary, length); + s->strstart = length; + s->block_start = (long)length; + + /* Insert all strings in the hash table (except for the last two bytes). + * s->lookahead stays null, so s->ins_h will be recomputed at the next + * call of fill_window. + */ + s->ins_h = s->window[0]; + UPDATE_HASH(s, s->ins_h, s->window[1]); + for (n = 0; n <= length - MIN_MATCH; n++) { + INSERT_STRING(s, n, hash_head); + } + if (hash_head) hash_head = 0; /* to make compiler happy */ + return Z_OK; +} + +/* ========================================================================= */ +int deflateReset (strm) + z_streamp strm; +{ + deflate_state *s; + + if (strm == Z_NULL || strm->state == Z_NULL || + strm->zalloc == Z_NULL || strm->zfree == Z_NULL) return Z_STREAM_ERROR; + + strm->total_in = strm->total_out = 0; + strm->msg = Z_NULL; /* use zfree if we ever allocate msg dynamically */ + strm->data_type = Z_UNKNOWN; + + s = (deflate_state *)strm->state; + s->pending = 0; + s->pending_out = s->pending_buf; + + if (s->noheader < 0) { + s->noheader = 0; /* was set to -1 by deflate(..., Z_FINISH); */ + } + s->status = s->noheader ? BUSY_STATE : INIT_STATE; + strm->adler = 1; + s->last_flush = Z_NO_FLUSH; + + _tr_init(s); + lm_init(s); + + return Z_OK; +} + +/* ========================================================================= */ +int deflateParams(strm, level, strategy) + z_streamp strm; + int level; + int strategy; +{ + deflate_state *s; + compress_func func; + int err = Z_OK; + + if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR; + s = (deflate_state *) strm->state; + + if (level == Z_DEFAULT_COMPRESSION) { + level = 6; + } + if (level < 0 || level > 9 || strategy < 0 || strategy > Z_HUFFMAN_ONLY) { + return Z_STREAM_ERROR; + } + func = configuration_table[s->level].func; + + if (func != configuration_table[level].func && strm->total_in != 0) { + /* Flush the last buffer: */ + err = deflate(strm, Z_PARTIAL_FLUSH); + } + if (s->level != level) { + s->level = level; + s->max_lazy_match = configuration_table[level].max_lazy; + s->good_match = configuration_table[level].good_length; + s->nice_match = configuration_table[level].nice_length; + s->max_chain_length = configuration_table[level].max_chain; + } + s->strategy = strategy; + return err; +} + +/* ========================================================================= + * Put a short in the pending buffer. The 16-bit value is put in MSB order. + * IN assertion: the stream state is correct and there is enough room in + * pending_buf. + */ +local void putShortMSB (s, b) + deflate_state *s; + uInt b; +{ + put_byte(s, (Byte)(b >> 8)); + put_byte(s, (Byte)(b & 0xff)); +} + +/* ========================================================================= + * Flush as much pending output as possible. All deflate() output goes + * through this function so some applications may wish to modify it + * to avoid allocating a large strm->next_out buffer and copying into it. + * (See also read_buf()). + */ +local void flush_pending(strm) + z_streamp strm; +{ + deflate_state *s = (deflate_state *) strm->state; + unsigned len = s->pending; + + if (len > strm->avail_out) len = strm->avail_out; + if (len == 0) return; + + if (strm->next_out != Z_NULL) { + zmemcpy(strm->next_out, s->pending_out, len); + strm->next_out += len; + } + s->pending_out += len; + strm->total_out += len; + strm->avail_out -= len; + s->pending -= len; + if (s->pending == 0) { + s->pending_out = s->pending_buf; + } +} + +/* ========================================================================= */ +int deflate (strm, flush) + z_streamp strm; + int flush; +{ + int old_flush; /* value of flush param for previous deflate call */ + deflate_state *s; + + if (strm == Z_NULL || strm->state == Z_NULL || + flush > Z_FINISH || flush < 0) { + return Z_STREAM_ERROR; + } + s = (deflate_state *) strm->state; + + if ((strm->next_in == Z_NULL && strm->avail_in != 0) || + (s->status == FINISH_STATE && flush != Z_FINISH)) { + ERR_RETURN(strm, Z_STREAM_ERROR); + } + if (strm->avail_out == 0) ERR_RETURN(strm, Z_BUF_ERROR); + + s->strm = strm; /* just in case */ + old_flush = s->last_flush; + s->last_flush = flush; + + /* Write the zlib header */ + if (s->status == INIT_STATE) { + + uInt header = (Z_DEFLATED + ((s->w_bits-8)<<4)) << 8; + uInt level_flags = (s->level-1) >> 1; + + if (level_flags > 3) level_flags = 3; + header |= (level_flags << 6); + if (s->strstart != 0) header |= PRESET_DICT; + header += 31 - (header % 31); + + s->status = BUSY_STATE; + putShortMSB(s, header); + + /* Save the adler32 of the preset dictionary: */ + if (s->strstart != 0) { + putShortMSB(s, (uInt)(strm->adler >> 16)); + putShortMSB(s, (uInt)(strm->adler & 0xffff)); + } + strm->adler = 1L; + } + + /* Flush as much pending output as possible */ + if (s->pending != 0) { + flush_pending(strm); + if (strm->avail_out == 0) { + /* Since avail_out is 0, deflate will be called again with + * more output space, but possibly with both pending and + * avail_in equal to zero. There won't be anything to do, + * but this is not an error situation so make sure we + * return OK instead of BUF_ERROR at next call of deflate: + */ + s->last_flush = -1; + return Z_OK; + } + + /* Make sure there is something to do and avoid duplicate consecutive + * flushes. For repeated and useless calls with Z_FINISH, we keep + * returning Z_STREAM_END instead of Z_BUFF_ERROR. + */ + } else if (strm->avail_in == 0 && flush <= old_flush && + flush != Z_FINISH) { + ERR_RETURN(strm, Z_BUF_ERROR); + } + + /* User must not provide more input after the first FINISH: */ + if (s->status == FINISH_STATE && strm->avail_in != 0) { + ERR_RETURN(strm, Z_BUF_ERROR); + } + + /* Start a new block or continue the current one. + */ + if (strm->avail_in != 0 || s->lookahead != 0 || + (flush != Z_NO_FLUSH && s->status != FINISH_STATE)) { + block_state bstate; + + bstate = (*(configuration_table[s->level].func))(s, flush); + + if (bstate == finish_started || bstate == finish_done) { + s->status = FINISH_STATE; + } + if (bstate == need_more || bstate == finish_started) { + if (strm->avail_out == 0) { + s->last_flush = -1; /* avoid BUF_ERROR next call, see above */ + } + return Z_OK; + /* If flush != Z_NO_FLUSH && avail_out == 0, the next call + * of deflate should use the same flush parameter to make sure + * that the flush is complete. So we don't have to output an + * empty block here, this will be done at next call. This also + * ensures that for a very small output buffer, we emit at most + * one empty block. + */ + } + if (bstate == block_done) { + if (flush == Z_PARTIAL_FLUSH) { + _tr_align(s); + } else if (flush == Z_PACKET_FLUSH) { + /* Output just the 3-bit `stored' block type value, + but not a zero length. */ + _tr_stored_type_only(s); + } else { /* FULL_FLUSH or SYNC_FLUSH */ + _tr_stored_block(s, (char*)0, 0L, 0); + /* For a full flush, this empty block will be recognized + * as a special marker by inflate_sync(). + */ + if (flush == Z_FULL_FLUSH) { + CLEAR_HASH(s); /* forget history */ + } + } + flush_pending(strm); + if (strm->avail_out == 0) { + s->last_flush = -1; /* avoid BUF_ERROR at next call, see above */ + return Z_OK; + } + } + } + Assert(strm->avail_out > 0, "bug2"); + + if (flush != Z_FINISH) return Z_OK; + if (s->noheader) return Z_STREAM_END; + + /* Write the zlib trailer (adler32) */ + putShortMSB(s, (uInt)(strm->adler >> 16)); + putShortMSB(s, (uInt)(strm->adler & 0xffff)); + flush_pending(strm); + /* If avail_out is zero, the application will call deflate again + * to flush the rest. + */ + s->noheader = -1; /* write the trailer only once! */ + return s->pending != 0 ? Z_OK : Z_STREAM_END; +} + +/* ========================================================================= */ +int deflateEnd (strm) + z_streamp strm; +{ + int status; + deflate_state *s; + + if (strm == Z_NULL || strm->state == Z_NULL) return Z_STREAM_ERROR; + s = (deflate_state *) strm->state; + + status = s->status; + if (status != INIT_STATE && status != BUSY_STATE && + status != FINISH_STATE) { + return Z_STREAM_ERROR; + } + + /* Deallocate in reverse order of allocations: */ + TRY_FREE(strm, s->pending_buf); + TRY_FREE(strm, s->head); + TRY_FREE(strm, s->prev); + TRY_FREE(strm, s->window); + + ZFREE(strm, s); + strm->state = Z_NULL; + + return status == BUSY_STATE ? Z_DATA_ERROR : Z_OK; +} + +/* ========================================================================= + * Copy the source state to the destination state. + */ +int deflateCopy (dest, source) + z_streamp dest; + z_streamp source; +{ + deflate_state *ds; + deflate_state *ss; + ushf *overlay; + + if (source == Z_NULL || dest == Z_NULL || source->state == Z_NULL) + return Z_STREAM_ERROR; + ss = (deflate_state *) source->state; + + *dest = *source; + + ds = (deflate_state *) ZALLOC(dest, 1, sizeof(deflate_state)); + if (ds == Z_NULL) return Z_MEM_ERROR; + dest->state = (struct internal_state FAR *) ds; + *ds = *ss; + ds->strm = dest; + + ds->window = (Bytef *) ZALLOC(dest, ds->w_size, 2*sizeof(Byte)); + ds->prev = (Posf *) ZALLOC(dest, ds->w_size, sizeof(Pos)); + ds->head = (Posf *) ZALLOC(dest, ds->hash_size, sizeof(Pos)); + overlay = (ushf *) ZALLOC(dest, ds->lit_bufsize, sizeof(ush)+2); + ds->pending_buf = (uchf *) overlay; + + if (ds->window == Z_NULL || ds->prev == Z_NULL || ds->head == Z_NULL || + ds->pending_buf == Z_NULL) { + deflateEnd (dest); + return Z_MEM_ERROR; + } + /* ??? following zmemcpy doesn't work for 16-bit MSDOS */ + zmemcpy(ds->window, ss->window, ds->w_size * 2 * sizeof(Byte)); + zmemcpy(ds->prev, ss->prev, ds->w_size * sizeof(Pos)); + zmemcpy(ds->head, ss->head, ds->hash_size * sizeof(Pos)); + zmemcpy(ds->pending_buf, ss->pending_buf, (uInt)ds->pending_buf_size); + + ds->pending_out = ds->pending_buf + (ss->pending_out - ss->pending_buf); + ds->d_buf = overlay + ds->lit_bufsize/sizeof(ush); + ds->l_buf = ds->pending_buf + (1+sizeof(ush))*ds->lit_bufsize; + + ds->l_desc.dyn_tree = ds->dyn_ltree; + ds->d_desc.dyn_tree = ds->dyn_dtree; + ds->bl_desc.dyn_tree = ds->bl_tree; + + return Z_OK; +} + +/* =========================================================================== + * Return the number of bytes of output which are immediately available + * for output from the decompressor. + */ +int deflateOutputPending (strm) + z_streamp strm; +{ + if (strm == Z_NULL || strm->state == Z_NULL) return 0; + + return ((deflate_state *)(strm->state))->pending; +} + +/* =========================================================================== + * Read a new buffer from the current input stream, update the adler32 + * and total number of bytes read. All deflate() input goes through + * this function so some applications may wish to modify it to avoid + * allocating a large strm->next_in buffer and copying from it. + * (See also flush_pending()). + */ +local int read_buf(strm, buf, size) + z_streamp strm; + charf *buf; + unsigned size; +{ + unsigned len = strm->avail_in; + + if (len > size) len = size; + if (len == 0) return 0; + + strm->avail_in -= len; + + if (!((deflate_state *)(strm->state))->noheader) { + strm->adler = adler32(strm->adler, strm->next_in, len); + } + zmemcpy(buf, strm->next_in, len); + strm->next_in += len; + strm->total_in += len; + + return (int)len; +} + +/* =========================================================================== + * Initialize the "longest match" routines for a new zlib stream + */ +local void lm_init (s) + deflate_state *s; +{ + s->window_size = (ulg)2L*s->w_size; + + CLEAR_HASH(s); + + /* Set the default configuration parameters: + */ + s->max_lazy_match = configuration_table[s->level].max_lazy; + s->good_match = configuration_table[s->level].good_length; + s->nice_match = configuration_table[s->level].nice_length; + s->max_chain_length = configuration_table[s->level].max_chain; + + s->strstart = 0; + s->block_start = 0L; + s->lookahead = 0; + s->match_length = s->prev_length = MIN_MATCH-1; + s->match_available = 0; + s->ins_h = 0; +#ifdef ASMV + match_init(); /* initialize the asm code */ +#endif +} + +/* =========================================================================== + * Set match_start to the longest match starting at the given string and + * return its length. Matches shorter or equal to prev_length are discarded, + * in which case the result is equal to prev_length and match_start is + * garbage. + * IN assertions: cur_match is the head of the hash chain for the current + * string (strstart) and its distance is <= MAX_DIST, and prev_length >= 1 + * OUT assertion: the match length is not greater than s->lookahead. + */ +#ifndef ASMV +/* For 80x86 and 680x0, an optimized version will be provided in match.asm or + * match.S. The code will be functionally equivalent. + */ +local uInt longest_match(s, cur_match) + deflate_state *s; + IPos cur_match; /* current match */ +{ + unsigned chain_length = s->max_chain_length;/* max hash chain length */ + register Bytef *scan = s->window + s->strstart; /* current string */ + register Bytef *match; /* matched string */ + register int len; /* length of current match */ + int best_len = s->prev_length; /* best match length so far */ + int nice_match = s->nice_match; /* stop if match long enough */ + IPos limit = s->strstart > (IPos)MAX_DIST(s) ? + s->strstart - (IPos)MAX_DIST(s) : NIL; + /* Stop when cur_match becomes <= limit. To simplify the code, + * we prevent matches with the string of window index 0. + */ + Posf *prev = s->prev; + uInt wmask = s->w_mask; + +#ifdef UNALIGNED_OK + /* Compare two bytes at a time. Note: this is not always beneficial. + * Try with and without -DUNALIGNED_OK to check. + */ + register Bytef *strend = s->window + s->strstart + MAX_MATCH - 1; + register ush scan_start = *(ushf*)scan; + register ush scan_end = *(ushf*)(scan+best_len-1); +#else + register Bytef *strend = s->window + s->strstart + MAX_MATCH; + register Byte scan_end1 = scan[best_len-1]; + register Byte scan_end = scan[best_len]; +#endif + + /* The code is optimized for HASH_BITS >= 8 and MAX_MATCH-2 multiple of 16. + * It is easy to get rid of this optimization if necessary. + */ + Assert(s->hash_bits >= 8 && MAX_MATCH == 258, "Code too clever"); + + /* Do not waste too much time if we already have a good match: */ + if (s->prev_length >= s->good_match) { + chain_length >>= 2; + } + /* Do not look for matches beyond the end of the input. This is necessary + * to make deflate deterministic. + */ + if ((uInt)nice_match > s->lookahead) nice_match = s->lookahead; + + Assert((ulg)s->strstart <= s->window_size-MIN_LOOKAHEAD, "need lookahead"); + + do { + Assert(cur_match < s->strstart, "no future"); + match = s->window + cur_match; + + /* Skip to next match if the match length cannot increase + * or if the match length is less than 2: + */ +#if (defined(UNALIGNED_OK) && MAX_MATCH == 258) + /* This code assumes sizeof(unsigned short) == 2. Do not use + * UNALIGNED_OK if your compiler uses a different size. + */ + if (*(ushf*)(match+best_len-1) != scan_end || + *(ushf*)match != scan_start) continue; + + /* It is not necessary to compare scan[2] and match[2] since they are + * always equal when the other bytes match, given that the hash keys + * are equal and that HASH_BITS >= 8. Compare 2 bytes at a time at + * strstart+3, +5, ... up to strstart+257. We check for insufficient + * lookahead only every 4th comparison; the 128th check will be made + * at strstart+257. If MAX_MATCH-2 is not a multiple of 8, it is + * necessary to put more guard bytes at the end of the window, or + * to check