Re: [PATCH v10 05/14] unwind_user/deferred: Add unwind cache
From: Steven Rostedt <rostedt@goodmis.org>
Date: 2025-06-18 15:33:58
Also in:
bpf, lkml
On Wed, 18 Jun 2025 16:13:45 +0200 Peter Zijlstra [off-list ref] wrote:
quoted
diff --git a/include/linux/entry-common.h b/include/linux/entry-common.h index f94f3fdf15fc..6e850c9d3f0c 100644 --- a/include/linux/entry-common.h +++ b/include/linux/entry-common.h@@ -12,6 +12,7 @@ #include <linux/resume_user_mode.h> #include <linux/tick.h> #include <linux/kmsan.h> +#include <linux/unwind_deferred.h> #include <asm/entry-common.h> #include <asm/syscall.h>@@ -362,6 +363,7 @@ static __always_inline void exit_to_user_mode(void) lockdep_hardirqs_on_prepare(); instrumentation_end(); + unwind_exit_to_user_mode();So I was expecting this to do the actual unwind, and was about to go yell this is the wrong place for that. But this is not that. Perhaps find a better name like: unwind_clear_cache() or so?
Sure. How about unwind_reset_info()? As it's not going to just clear the cache but also reset the trace info (like the timestamp and such).
quoted
user_enter_irqoff(); arch_exit_to_user_mode(); lockdep_hardirqs_on(CALLER_ADDR0);quoted
diff --git a/include/linux/unwind_deferred_types.h b/include/linux/unwind_deferred_types.h index aa32db574e43..db5b54b18828 100644 --- a/include/linux/unwind_deferred_types.h +++ b/include/linux/unwind_deferred_types.h@@ -2,8 +2,13 @@ #ifndef _LINUX_UNWIND_USER_DEFERRED_TYPES_H #define _LINUX_UNWIND_USER_DEFERRED_TYPES_H +struct unwind_cache { + unsigned int nr_entries; + unsigned long entries[]; +}; + struct unwind_task_info { - unsigned long *entries; + struct unwind_cache *cache; }; #endif /* _LINUX_UNWIND_USER_DEFERRED_TYPES_H */diff --git a/kernel/unwind/deferred.c b/kernel/unwind/deferred.c index 0bafb95e6336..e3913781c8c6 100644 --- a/kernel/unwind/deferred.c +++ b/kernel/unwind/deferred.c@@ -24,6 +24,7 @@ int unwind_deferred_trace(struct unwind_stacktrace *trace) { struct unwind_task_info *info = ¤t->unwind_info; + struct unwind_cache *cache; /* Should always be called from faultable context */ might_fault();@@ -31,17 +32,30 @@ int unwind_deferred_trace(struct unwind_stacktrace *trace) if (current->flags & PF_EXITING) return -EINVAL; - if (!info->entries) { - info->entries = kmalloc_array(UNWIND_MAX_ENTRIES, sizeof(long), - GFP_KERNEL); - if (!info->entries) + if (!info->cache) { + info->cache = kzalloc(struct_size(cache, entries, UNWIND_MAX_ENTRIES), + GFP_KERNEL);And now you're one 'long' larger than a page. Surely that's a crap size for an allocator?
Bah, Ingo suggested to put the counter in the allocation and I didn't think about the size going over the page. Good catch! Since it can make one per task, it may be good to make this into a kmemcache. -- Steve