Remove default GFP_NOFS allocations

As originally described in commit 82b8c8fa64
this was done to prevent certain deadlocks from occuring in the system.
However, as suspected the price for doing this proved to be too high.
The VM is having a hard time effectively reclaiming memory thus we are
reverting this change.

However, we still need to fundamentally handle the issue.  Under
Solaris the KM_PUSHPAGE mask is used commonly in I/O paths to ensure
a memory allocations will succeed.  We leverage this fact and redefine
KM_PUSHPAGE to include GFP_NOFS.  This ensures that in these common
I/O path we don't trigger additional reclaim.  This minimizes the
change to the Solaris code.
This commit is contained in:
Brian Behlendorf 2011-03-19 13:49:14 -07:00
parent 181a9b8998
commit cb255ae572
1 changed files with 6 additions and 7 deletions

View File

@ -41,13 +41,12 @@
/* /*
* Memory allocation interfaces * Memory allocation interfaces
*/ */
#define KM_SLEEP GFP_NOFS #define KM_SLEEP GFP_KERNEL /* Can sleep, never fails */
#define KM_NOSLEEP GFP_ATOMIC #define KM_NOSLEEP GFP_ATOMIC /* Can not sleep, may fail */
#undef KM_PANIC /* No linux analog */ #define KM_PUSHPAGE (GFP_NOFS | __GFP_HIGH) /* Use reserved memory */
#define KM_PUSHPAGE (KM_SLEEP | __GFP_HIGH) #define KM_NODEBUG __GFP_NOWARN /* Suppress warnings */
#define KM_VMFLAGS GFP_LEVEL_MASK #define KM_FLAGS __GFP_BITS_MASK
#define KM_FLAGS __GFP_BITS_MASK #define KM_VMFLAGS GFP_LEVEL_MASK
#define KM_NODEBUG __GFP_NOWARN
/* /*
* Used internally, the kernel does not need to support this flag * Used internally, the kernel does not need to support this flag