kdb: Make memory allocations more robust
PATCH 1/1 kdb: Make memory allocations more robust
Currently kdb uses in_interrupt() to determine whether it's library code has been called from the kgdb trap handler or from a saner calling context such as driver init. This approach is broken because in_interrupt() alone isn't able to determine kgdb trap handler entry via normal task context such as [1].
We can improve this by adding check for in_dbg_master() which explicitly determines if we are running in debugger context. Also, use in_atomic() instead of in_interrupt() as the former is more appropriate to know atomic context and moreover the later one is deprecated.
[1] $ echo g > /proc/sysrq-trigger
--- kernel/debug/kdb/kdb_private.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/debug/kdb/kdb_private.h +1 -1
@@ -231,7 +231,7 @@ extern struct task_struct *kdb_curr_task(int);#define kdb_task_has_cpu (p) (task_curr(p)) -#define GFP_KDB (in_interrupt() ? GFP_ATOMIC : GFP_KERNEL) +#define GFP_KDB (in_atomic() || in_dbg_master() ? GFP_ATOMIC : GFP_KERNEL)extern void *debug_kmalloc (size_t size,gfp_t flags);extern void debug_kfree (void *);
Signature · 1 line
2.7.4
Quoted 10 lines · Sumit Garg
Currently kdb uses
in_interrupt()to determine whether it's library code has been called from the kgdb trap handler or from a saner calling context such as driver init. This approach is broken becausein_interrupt()alone isn't able to determine kgdb trap handler entry via normal task context such as [1].We can improve this by adding check for
in_dbg_master()which explicitly determines if we are running in debugger context. Also, usein_atomic()instead ofin_interrupt()as the former is more appropriate to know atomic context and moreover the later one is deprecated.
Why do we need the in_atomic() here? Or put another way, why isn't in_dbg_master() sufficient?
Daniel.
Quoted 24 lines
[1] $ echo g >
/proc/sysrq-triggerSigned-off-by Sumit Garg <[email protected]>--- kernel/debug/kdb/kdb_private.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)kernel/debug/kdb/kdb_private.h +1
-1@@ -231,7 +231,7 @@ extern struct task_struct *kdb_curr_task(int);#define kdb_task_has_cpu (p) (task_curr(p)) -#define GFP_KDB (in_interrupt() ? GFP_ATOMIC : GFP_KERNEL) +#define GFP_KDB (in_atomic() || in_dbg_master() ? GFP_ATOMIC : GFP_KERNEL)extern void *debug_kmalloc (size_t size,gfp_t flags);extern void debug_kfree (void *);Signature · 1 line
2.7.4
Quoted 8 lines · Daniel Thompson
Quoted 6 lines · Sumit Garg
Currently kdb uses
in_interrupt()to determine whether it's library code has been called from the kgdb trap handler or from a saner calling context such as driver init. This approach is broken becausein_interrupt()alone isn't able to determine kgdb trap handler entry via normal task context such as [1].
Quoted 8 lines
Quoted 4 lines
We can improve this by adding check for
in_dbg_master()which explicitly determines if we are running in debugger context. Also, usein_atomic()instead ofin_interrupt()as the former is more appropriate to know atomic context and moreover the later one is deprecated.Why do we need the
in_atomic()here? Or put another way, why isn'tin_dbg_master()sufficient?
Yes, you are right in_atomic() is redundant after looking at usage of GFP_KDB. Will get rid of it in v2.
-Sumit
Quoted 28 lines
Daniel.
Quoted 24 lines
[1] $ echo g >
/proc/sysrq-triggerSigned-off-by Sumit Garg <[email protected]>--- kernel/debug/kdb/kdb_private.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)kernel/debug/kdb/kdb_private.h +1
-2@@ -231,7 +231,7 @@ extern struct task_struct *kdb_curr_task(int);#define kdb_task_has_cpu (p) (task_curr(p)) -#define GFP_KDB (in_interrupt() ? GFP_ATOMIC : GFP_KERNEL) +#define GFP_KDB (in_atomic() || in_dbg_master() ? GFP_ATOMIC : GFP_KERNEL)extern void *debug_kmalloc (size_t size,gfp_t flags);extern void debug_kfree (void *); --2.7.4