Skip to contentSkip to conversation

powerpc/iommu: Implement IOMMU pools to improve multiqueue adapter performance

bug report Dan Carpenter 1 participant 1 message Aug 14, 2019

lore
Share & tools
lore
#0 lore raw

[ Ancient code. The warning is correct but the bug seems harmless.
-- dan ]

Hello Anton Blanchard,

The patch b4c3a8729ae5: "powerpc/iommu: Implement IOMMU pools to improve multiqueue adapter performance" from Jun 7, 2012, leads to the following static checker warning:

	arch/powerpc/kernel/iommu.c:377 get_pool()
	warn: array off by one? '*tbl->pools + pool_nr'

arch/powerpc/kernel/iommu.c
364 static struct iommu_pool *get_pool(struct iommu_table *tbl,
365 unsigned long entry)
366 {
367 struct iommu_pool *p;
368 unsigned long largepool_start = tbl->large_pool.start;
369
370 /* The large pool is the last pool at the top of the table */
371 if (entry >= largepool_start) {
372 p = &tbl->large_pool;
373 } else {
374 unsigned int pool_nr = entry / tbl->poolsize;
375

   376                  BUG_ON(pool_nr > tbl->nr_pools);
                                       ^

This should be ">=". The tbl->nr_pools value is either 1 or IOMMU_NR_POOLS and the tbl->pools[] array has IOMMU_NR_POOLS elements.

   377                  p = &tbl->pools[pool_nr];
   378          }
   379  
   380          return p;
   381  }

regards,
dan carpenter