[IB] mthca: Fix doorbell record resource leak
If we allocate a bunch of doorbell records and then free them, we'll end up with completely empty pages, which we then free. However, when we come back to allocate more doorbell pages, we have to reallocate those empty pages rather than always trying to take a slot that we've never used. If we don't, we eventually use up every slot and fail to allocate a doorbell record, even though we have plenty of free space. Signed-off-by: Roland Dreier <rolandd@cisco.com>
This commit is contained in:
parent
8ddec7460d
commit
018771f435
|
@ -529,12 +529,25 @@ int mthca_alloc_db(struct mthca_dev *dev, int type, u32 qn, __be32 **db)
|
|||
goto found;
|
||||
}
|
||||
|
||||
for (i = start; i != end; i += dir)
|
||||
if (!dev->db_tab->page[i].db_rec) {
|
||||
page = dev->db_tab->page + i;
|
||||
goto alloc;
|
||||
}
|
||||
|
||||
if (dev->db_tab->max_group1 >= dev->db_tab->min_group2 - 1) {
|
||||
ret = -ENOMEM;
|
||||
goto out;
|
||||
}
|
||||
|
||||
if (group == 0)
|
||||
++dev->db_tab->max_group1;
|
||||
else
|
||||
--dev->db_tab->min_group2;
|
||||
|
||||
page = dev->db_tab->page + end;
|
||||
|
||||
alloc:
|
||||
page->db_rec = dma_alloc_coherent(&dev->pdev->dev, 4096,
|
||||
&page->mapping, GFP_KERNEL);
|
||||
if (!page->db_rec) {
|
||||
|
@ -554,10 +567,6 @@ int mthca_alloc_db(struct mthca_dev *dev, int type, u32 qn, __be32 **db)
|
|||
}
|
||||
|
||||
bitmap_zero(page->used, MTHCA_DB_REC_PER_PAGE);
|
||||
if (group == 0)
|
||||
++dev->db_tab->max_group1;
|
||||
else
|
||||
--dev->db_tab->min_group2;
|
||||
|
||||
found:
|
||||
j = find_first_zero_bit(page->used, MTHCA_DB_REC_PER_PAGE);
|
||||
|
|
Loading…
Reference in a new issue