Andi Kleen wrote: > Jason Wessel <jason.wessel@xxxxxxxxxxxxx> writes: > > I remember going with kaos through all the code > outside kdb/ in his own patch and for nearly all hooks > outside we found some way to eliminate them. > > I think a lot of this is here too. If this has been solved before somewhere, please point me in that direction. There are whole lot few changes in this series vs original kdb outside of the kdb core area. I already split all the files out of the large kdb patch that are modified to support the kdb for the next time I post this series. I included the updated version of the non-kdb files patch in this mail. > >> >> +#ifdef CONFIG_KGDB_KDB >> +/* Like meminfo_proc_show() but without the locks and using kdb_printf() */ >> +void kdb_meminfo_proc_show(void) > > Are there even any locks in meminfo_proc_show()? I don't see any on a quick look. > Ah or is that only for swap_info? That could be a flag or perhaps that > access can be even made lockless (it looks like it could) Yup it definitely needs to be lockless but can be controlled with a flag. > > I guess a better way would be to have a kdb specific seq file implementation > and then just use the normal function, instead of copying everything. > Sure, this seems a whole lot more reasonable than the bulk copy and maintenance of the function specifically for kdb. I went ahead and implemented a (kdb_seq_file *) in the kdb core so I could just directly call the _meminfo_proc_show() without the locks. >> void get_vmalloc_info(struct vmalloc_info *vmi) >> { >> struct vm_struct *vma; >> unsigned long free_area_size; >> unsigned long prev_end; >> +#ifdef CONFIG_KGDB_KDB >> + int get_lock = !KDB_IS_RUNNING(); >> +#else >> +#define get_lock 1 >> +#endif >> + > > A standard way to do such would be a __get_vmalloc_info with the > lock in the caller That particular #ifdef is gone now, and traded for a flag passed in with the function. Between the changes that you recommended and that Peter recommended, the scope of changes to files out side the kdb area has been reduced by several hundred lines. Thanks, Jason.
From: Jason Wessel <jason.wessel@xxxxxxxxxxxxx> Subject: [PATCH] kdb: core for kgdb back end (2 of 2) This patch contains the hooks and instrumentation into kernel which live outside the kernel/debug directory, which the kdb core will call to run commands like lsmod, dmesg, bt etc... CC: mort@xxxxxxx CC: linux-arch@xxxxxxxxxxxxxxx Signed-off-by: Jason Wessel <jason.wessel@xxxxxxxxxxxxx> --- arch/arm/include/asm/kmap_types.h | 1 arch/powerpc/include/asm/kmap_types.h | 1 fs/proc/internal.h | 2 - fs/proc/meminfo.c | 15 +++++++++--- fs/proc/mmu.c | 8 ++++-- include/asm-generic/kmap_types.h | 3 +- include/linux/swap.h | 2 + init/main.c | 6 ++++ kernel/kallsyms.c | 21 +++++++++++++++++ kernel/module.c | 4 +++ kernel/printk.c | 16 ++++++++++++ kernel/sched.c | 7 ++++- kernel/signal.c | 42 ++++++++++++++++++++++++++++++++++ mm/swapfile.c | 10 ++++++-- 14 files changed, 126 insertions(+), 12 deletions(-) --- a/arch/arm/include/asm/kmap_types.h +++ b/arch/arm/include/asm/kmap_types.h @@ -19,6 +19,7 @@ enum km_type { KM_SOFTIRQ0, KM_SOFTIRQ1, KM_L2_CACHE, + KM_KDB, KM_TYPE_NR }; --- a/arch/powerpc/include/asm/kmap_types.h +++ b/arch/powerpc/include/asm/kmap_types.h @@ -26,6 +26,7 @@ enum km_type { KM_SOFTIRQ1, KM_PPC_SYNC_PAGE, KM_PPC_SYNC_ICACHE, + KM_KDB, KM_TYPE_NR }; --- a/fs/proc/meminfo.c +++ b/fs/proc/meminfo.c @@ -10,6 +10,7 @@ #include <linux/seq_file.h> #include <linux/swap.h> #include <linux/vmstat.h> +#include <linux/kdb.h> #include <asm/atomic.h> #include <asm/page.h> #include <asm/pgtable.h> @@ -19,7 +20,7 @@ void __attribute__((weak)) arch_report_m { } -static int meminfo_proc_show(struct seq_file *m, void *v) +int _meminfo_proc_show(struct seq_file *m, void *v, int lock) { struct sysinfo i; unsigned long committed; @@ -34,7 +35,10 @@ static int meminfo_proc_show(struct seq_ */ #define K(x) ((x) << (PAGE_SHIFT - 10)) si_meminfo(&i); - si_swapinfo(&i); + if (lock) + si_swapinfo(&i); + else + __si_swapinfo(&i); committed = percpu_counter_read_positive(&vm_committed_as); allowed = ((totalram_pages - hugetlb_total_pages()) * sysctl_overcommit_ratio / 100) + total_swap_pages; @@ -44,7 +48,7 @@ static int meminfo_proc_show(struct seq_ if (cached < 0) cached = 0; - get_vmalloc_info(&vmi); + get_vmalloc_info(&vmi, lock); for (lru = LRU_BASE; lru < NR_LRU_LISTS; lru++) pages[lru] = global_page_state(NR_LRU_BASE + lru); @@ -161,6 +165,11 @@ static int meminfo_proc_show(struct seq_ #undef K } +static int meminfo_proc_show(struct seq_file *m, void *v) +{ + return _meminfo_proc_show(m, v, 1); +} + static int meminfo_proc_open(struct inode *inode, struct file *file) { return single_open(file, meminfo_proc_show, NULL); --- a/fs/proc/mmu.c +++ b/fs/proc/mmu.c @@ -14,7 +14,7 @@ #include <asm/pgtable.h> #include "internal.h" -void get_vmalloc_info(struct vmalloc_info *vmi) +void get_vmalloc_info(struct vmalloc_info *vmi, int lock) { struct vm_struct *vma; unsigned long free_area_size; @@ -30,7 +30,8 @@ void get_vmalloc_info(struct vmalloc_inf prev_end = VMALLOC_START; - read_lock(&vmlist_lock); + if (lock) + read_lock(&vmlist_lock); for (vma = vmlist; vma; vma = vma->next) { unsigned long addr = (unsigned long) vma->addr; @@ -55,6 +56,7 @@ void get_vmalloc_info(struct vmalloc_inf if (VMALLOC_END - prev_end > vmi->largest_chunk) vmi->largest_chunk = VMALLOC_END - prev_end; - read_unlock(&vmlist_lock); + if (lock) + read_unlock(&vmlist_lock); } } --- a/include/asm-generic/kmap_types.h +++ b/include/asm-generic/kmap_types.h @@ -28,7 +28,8 @@ KMAP_D(15) KM_UML_USERCOPY, KMAP_D(16) KM_IRQ_PTE, KMAP_D(17) KM_NMI, KMAP_D(18) KM_NMI_PTE, -KMAP_D(19) KM_TYPE_NR +KMAP_D(19) KM_KDB, +KMAP_D(20) KM_TYPE_NR }; #undef KMAP_D --- a/include/linux/swap.h +++ b/include/linux/swap.h @@ -312,6 +312,7 @@ extern struct page *swapin_readahead(swp /* linux/mm/swapfile.c */ extern long nr_swap_pages; extern long total_swap_pages; +extern void __si_swapinfo(struct sysinfo *); extern void si_swapinfo(struct sysinfo *); extern swp_entry_t get_swap_page(void); extern swp_entry_t get_swap_page_of_type(int); @@ -377,6 +378,7 @@ static inline void mem_cgroup_uncharge_s #define si_swapinfo(val) \ do { (val)->freeswap = (val)->totalswap = 0; } while (0) +#define __si_swapinfo(val) si_swapinfo(val) /* only sparc can not include linux/pagemap.h in this file * so leave page_cache_release and release_pages undeclared... */ #define free_page_and_swap_cache(page) \ --- a/init/main.c +++ b/init/main.c @@ -63,6 +63,7 @@ #include <linux/sched.h> #include <linux/signal.h> #include <linux/idr.h> +#include <linux/kdb.h> #include <linux/ftrace.h> #include <linux/async.h> #include <linux/kmemcheck.h> @@ -647,6 +648,11 @@ asmlinkage void __init start_kernel(void calibrate_delay(); pidmap_init(); anon_vma_init(); + +#ifdef CONFIG_KGDB_KDB + kdb_init(); +#endif /* CONFIG_KGDB_KDB */ + #ifdef CONFIG_X86 if (efi_enabled) efi_enter_virtual_mode(); --- a/kernel/kallsyms.c +++ b/kernel/kallsyms.c @@ -16,6 +16,7 @@ #include <linux/init.h> #include <linux/seq_file.h> #include <linux/fs.h> +#include <linux/kdb.h> #include <linux/err.h> #include <linux/proc_fs.h> #include <linux/sched.h> /* for cond_resched */ @@ -515,6 +516,26 @@ static int kallsyms_open(struct inode *i return ret; } +#ifdef CONFIG_KGDB_KDB +const char *kdb_walk_kallsyms(loff_t *pos) +{ + static struct kallsym_iter kdb_walk_kallsyms_iter; + if (*pos == 0) { + memset(&kdb_walk_kallsyms_iter, 0, + sizeof(kdb_walk_kallsyms_iter)); + reset_iter(&kdb_walk_kallsyms_iter, 0); + } + while (1) { + if (!update_iter(&kdb_walk_kallsyms_iter, *pos)) + return NULL; + ++*pos; + /* Some debugging symbols have no name. Ignore them. */ + if (kdb_walk_kallsyms_iter.name[0]) + return kdb_walk_kallsyms_iter.name; + } +} +#endif /* CONFIG_KGDB_KDB */ + static const struct file_operations kallsyms_operations = { .open = kallsyms_open, .read = seq_read, --- a/kernel/module.c +++ b/kernel/module.c @@ -79,6 +79,10 @@ EXPORT_TRACEPOINT_SYMBOL(module_get); DEFINE_MUTEX(module_mutex); EXPORT_SYMBOL_GPL(module_mutex); static LIST_HEAD(modules); +#ifdef CONFIG_KGDB_KDB +struct list_head *kdb_modules = &modules; /* kdb needs the list of modules */ +#endif /* CONFIG_KGDB_KDB */ + /* Block module loading/unloading? */ int modules_disabled = 0; --- a/kernel/printk.c +++ b/kernel/printk.c @@ -420,6 +420,22 @@ SYSCALL_DEFINE3(syslog, int, type, char return do_syslog(type, buf, len); } +#ifdef CONFIG_KGDB_KDB +/* kdb dmesg command needs access to the syslog buffer. do_syslog() + * uses locks so it cannot be used during debugging. Just tell kdb + * where the start and end of the physical and logical logs are. This + * is equivalent to do_syslog(3). + */ +void kdb_syslog_data(char *syslog_data[4]) +{ + syslog_data[0] = log_buf; + syslog_data[1] = log_buf + log_buf_len; + syslog_data[2] = log_buf + log_end - + (logged_chars < log_buf_len ? logged_chars : log_buf_len); + syslog_data[3] = log_buf + log_end; +} +#endif /* CONFIG_KGDB_KDB */ + /* * Call the console drivers on a range of log_buf */ --- a/kernel/sched.c +++ b/kernel/sched.c @@ -9784,9 +9784,9 @@ void normalize_rt_tasks(void) #endif /* CONFIG_MAGIC_SYSRQ */ -#ifdef CONFIG_IA64 +#if defined(CONFIG_IA64) || defined(CONFIG_KGDB_KDB) /* - * These functions are only useful for the IA64 MCA handling. + * These functions are only useful for the IA64 MCA handling, or kdb. * * They can only be called when the whole system has been * stopped - every CPU needs to be quiescent, and no scheduling @@ -9806,6 +9806,9 @@ struct task_struct *curr_task(int cpu) return cpu_curr(cpu); } +#endif /* defined(CONFIG_IA64) || defined(CONFIG_KGDB_KDB) */ + +#ifdef CONFIG_IA64 /** * set_curr_task - set the current task for a given cpu. * @cpu: the processor in question. --- a/kernel/signal.c +++ b/kernel/signal.c @@ -2717,3 +2717,45 @@ void __init signals_init(void) { sigqueue_cachep = KMEM_CACHE(sigqueue, SLAB_PANIC); } + +#ifdef CONFIG_KGDB_KDB +#include <linux/kdb.h> +/* + * kdb_send_sig_info - Allows kdb to send signals without exposing + * signal internals. This function checks if the required locks are + * available before calling the main signal code, to avoid kdb + * deadlocks. + */ +void +kdb_send_sig_info(struct task_struct *t, struct siginfo *info, int seqno) +{ + static struct task_struct *kdb_prev_t; + static int kdb_prev_seqno; + int sig, new_t; + if (!spin_trylock(&t->sighand->siglock)) { + kdb_printf("Can't do kill command now.\n" + "The sigmask lock is held somewhere else in " + "kernel, try again later\n"); + return; + } + spin_unlock(&t->sighand->siglock); + new_t = kdb_prev_t != t || kdb_prev_seqno != seqno; + kdb_prev_t = t; + kdb_prev_seqno = seqno; + if (t->state != TASK_RUNNING && new_t) { + kdb_printf("Process is not RUNNING, sending a signal from " + "kdb risks deadlock\n" + "on the run queue locks. " + "The signal has _not_ been sent.\n" + "Reissue the kill command if you want to risk " + "the deadlock.\n"); + return; + } + sig = info->si_signo; + if (send_sig_info(sig, info, t)) + kdb_printf("Fail to deliver Signal %d to process %d.\n", + sig, t->pid); + else + kdb_printf("Signal %d is sent to process %d.\n", sig, t->pid); +} +#endif /* CONFIG_KGDB_KDB */ --- a/mm/swapfile.c +++ b/mm/swapfile.c @@ -13,6 +13,7 @@ #include <linux/swap.h> #include <linux/vmalloc.h> #include <linux/pagemap.h> +#include <linux/kdb.h> #include <linux/namei.h> #include <linux/shm.h> #include <linux/blkdev.h> @@ -2056,12 +2057,11 @@ out: return error; } -void si_swapinfo(struct sysinfo *val) +void __si_swapinfo(struct sysinfo *val) { unsigned int type; unsigned long nr_to_be_unused = 0; - spin_lock(&swap_lock); for (type = 0; type < nr_swapfiles; type++) { struct swap_info_struct *si = swap_info[type]; @@ -2070,6 +2070,12 @@ void si_swapinfo(struct sysinfo *val) } val->freeswap = nr_swap_pages + nr_to_be_unused; val->totalswap = total_swap_pages + nr_to_be_unused; +} + +void si_swapinfo(struct sysinfo *val) +{ + spin_lock(&swap_lock); + __si_swapinfo(val); spin_unlock(&swap_lock); } --- a/fs/proc/internal.h +++ b/fs/proc/internal.h @@ -32,7 +32,7 @@ extern struct mm_struct *mm_for_maps(str #ifdef CONFIG_MMU #define VMALLOC_TOTAL (VMALLOC_END - VMALLOC_START) -extern void get_vmalloc_info(struct vmalloc_info *vmi); +extern void get_vmalloc_info(struct vmalloc_info *vmi, int lock); #else #define VMALLOC_TOTAL 0UL