The patch titled
Subject: mm/dmapool.c: fix boundary comparison
has been added to the -mm tree. Its filename is
dmapool-fix-boundary-comparison.patch
This patch should soon appear at
http://ozlabs.org/~akpm/mmots/broken-out/dmapool-fix-boundary-comparison.patch
and later at
http://ozlabs.org/~akpm/mmotm/broken-out/dmapool-fix-boundary-comparison.patch
Before you just go and hit "reply", please:
a) Consider who else should be cc'ed
b) Prefer to cc a suitable mailing list as well
c) Ideally: find the original patch on the mailing list and do a
reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/process/submit-checklist.rst when testing your code ***
The -mm tree is included into linux-next and is updated
there every 3-4 working days
------------------------------------------------------
From: Tony Battersby <tonyb@xxxxxxxxxxxxxxx>
Subject: mm/dmapool.c: fix boundary comparison
Patch series "mpt3sas and dmapool scalability", v4.
drivers/scsi/mpt3sas is running into a scalability problem with the
kernel's DMA pool implementation. With a LSI/Broadcom SAS 9300-8i 12Gb/s
HBA and max_sgl_entries=256, during modprobe, mpt3sas does the equivalent
of:
chain_dma_pool = dma_pool_create(size = 128);
for (i = 0; i < 373959; i++)
{
dma_addr[i] = dma_pool_alloc(chain_dma_pool);
}
And at rmmod, system shutdown, or system reboot, mpt3sas does the
equivalent of:
for (i = 0; i < 373959; i++)
{
dma_pool_free(chain_dma_pool, dma_addr[i]);
}
dma_pool_destroy(chain_dma_pool);
With this usage, both dma_pool_alloc() and dma_pool_free() exhibit O(n^2)
complexity, although dma_pool_free() is much worse due to implementation
details. On my system, the dma_pool_free() loop above takes about 9
seconds to run. Note that the problem was even worse before commit
74522a92bbf0 ("scsi: mpt3sas: Optimize I/O memory consumption in
driver."), where the dma_pool_free() loop could take ~30 seconds.
mpt3sas also has some other DMA pools, but chain_dma_pool is the only
one with so many allocations:
cat /sys/devices/pci0000:80/0000:80:07.0/0000:85:00.0/pools
(manually cleaned up column alignment)
poolinfo - 0.1
reply_post_free_array pool 1 21 192 1
reply_free pool 1 1 41728 1
reply pool 1 1 1335296 1
sense pool 1 1 970272 1
chain pool 373959 386048 128 12064
reply_post_free pool 12 12 166528 12
The patches in this series improve the scalability of the DMA pool
implementation, which significantly reduces the running time of the DMA
alloc/free loops. With the patches applied, "modprobe mpt3sas", "rmmod
mpt3sas", and system shutdown/reboot with mpt3sas loaded are significantly
faster. Here are some benchmarks (of DMA alloc/free only, not the entire
modprobe/rmmod):
This patch (of 9):
Fix the boundary comparison when constructing the list of free blocks for
the case that 'size' is a power of two. Since 'boundary' is also a power
of two, that would make 'boundary' a multiple of 'size', in which case a
single block would never cross the boundary. This bug would cause some of
the allocated memory to be wasted (but not leaked).
Example:
size = 512
boundary = 2048
allocation = 4096
Address range
0 - 511
512 - 1023
1024 - 1535
1536 - 2047 *
2048 - 2559
2560 - 3071
3072 - 3583
3584 - 4095 *
Prior to this fix, the address ranges marked with "*" would not have been
used even though they didn't cross the given boundary.
Link: http://lkml.kernel.org/r/acce3a38-9930-349d-5299-95d2aa5c47e4@xxxxxxxxxxxxxxx
Fixes: e34f44b3517f ("pool: Improve memory usage for devices which can't cross boundaries")
Signed-off-by: Tony Battersby <tonyb@xxxxxxxxxxxxxxx>
Acked-by: Matthew Wilcox <willy@xxxxxxxxxxxxx>
Cc: Christoph Hellwig <hch@xxxxxx>
Cc: Marek Szyprowski <m.szyprowski@xxxxxxxxxxx>
Cc: Andy Shevchenko <andy.shevchenko@xxxxxxxxx>
Cc: John Garry <john.garry@xxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
mm/dmapool.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
--- a/mm/dmapool.c~dmapool-fix-boundary-comparison
+++ a/mm/dmapool.c
@@ -210,7 +210,7 @@ static void pool_initialise_page(struct
do {
unsigned int next = offset + pool->size;
- if (unlikely((next + pool->size) >= next_boundary)) {
+ if (unlikely((next + pool->size) > next_boundary)) {
next = next_boundary;
next_boundary += pool->boundary;
}
_
Patches currently in -mm which might be from tonyb@xxxxxxxxxxxxxxx are
dmapool-fix-boundary-comparison.patch
dmapool-remove-checks-for-dev-==-null.patch
dmapool-cleanup-dma_pool_destroy.patch
dmapool-improve-scalability-of-dma_pool_alloc.patch
dmapool-rename-fields-in-dma_page.patch
dmapool-improve-scalability-of-dma_pool_free.patch
dmapool-cleanup-integer-types.patch
dmapool-improve-accuracy-of-debug-statistics.patch
dmapool-debug-prevent-endless-loop-in-case-of-corruption.patch
