| Message ID | 1454042578-13933-1-git-send-email-shankerd@codeaurora.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Thu, 28 Jan 2016, Shanker Donthineni wrote: > @@ -807,9 +810,10 @@ static void its_free_tables(struct its_node *its) > int i; > > for (i = 0; i < GITS_BASER_NR_REGS; i++) { > - if (its->tables[i]) { > - free_page((unsigned long)its->tables[i]); > - its->tables[i] = NULL; > + if (its->tables[i].base) { > + free_pages((unsigned long)its->tables[i].base, > + get_order(its->tables[i].size)); > + its->tables[i].base = NULL; > } > } > } > @@ -880,6 +884,7 @@ retry_alloc_baser: > if (alloc_pages > GITS_BASER_PAGES_MAX) { > alloc_pages = GITS_BASER_PAGES_MAX; > order = get_order(GITS_BASER_PAGES_MAX * psz); > + alloc_size = (1 << order) * PAGE_SIZE; Why don't you record the order instead of converting back and forth ? Thanks, tglx
On 01/29/2016 02:30 AM, Thomas Gleixner wrote: > On Thu, 28 Jan 2016, Shanker Donthineni wrote: >> @@ -807,9 +810,10 @@ static void its_free_tables(struct its_node *its) >> int i; >> >> for (i = 0; i < GITS_BASER_NR_REGS; i++) { >> - if (its->tables[i]) { >> - free_page((unsigned long)its->tables[i]); >> - its->tables[i] = NULL; >> + if (its->tables[i].base) { >> + free_pages((unsigned long)its->tables[i].base, >> + get_order(its->tables[i].size)); >> + its->tables[i].base = NULL; >> } >> } >> } >> @@ -880,6 +884,7 @@ retry_alloc_baser: >> if (alloc_pages > GITS_BASER_PAGES_MAX) { >> alloc_pages = GITS_BASER_PAGES_MAX; >> order = get_order(GITS_BASER_PAGES_MAX * psz); >> + alloc_size = (1 << order) * PAGE_SIZE; > Why don't you record the order instead of converting back and forth ? I can use page order information to fix memory leak and I will post v3 patch with your suggestion. We have another coding BUG which is related to not refreshing alloc_size whenever order changes. Because we are not updating alloc_size variable here, later part of the code logic uses incorrect alloc_size value in two places as shown below. Code snippet-1: if (!shr) { cache = GITS_BASER_nC; __flush_dcache_area(base, alloc_size); } Code snippet-2: pr_info("ITS: allocated %d %s @%lx (psz %dK, shr %d)\n", (int)(alloc_size / entry_size), its_base_type_string[type], (unsigned long)virt_to_phys(base), psz / SZ_1K, (int)shr >> GITS_BASER_SHAREABILITY_SHIFT); How do you suggest I fix the second problem? Should I create another patch or include in v3 patch? > Thanks, > > tglx > > _______________________________________________ > linux-arm-kernel mailing list > linux-arm-kernel@lists.infradead.org > http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c index 3447549..fc1a34f 100644 --- a/drivers/irqchip/irq-gic-v3-its.c +++ b/drivers/irqchip/irq-gic-v3-its.c @@ -66,7 +66,10 @@ struct its_node { unsigned long phys_base; struct its_cmd_block *cmd_base; struct its_cmd_block *cmd_write; - void *tables[GITS_BASER_NR_REGS]; + struct { + void *base; + u32 size; + } tables[GITS_BASER_NR_REGS]; struct its_collection *collections; struct list_head its_device_list; u64 flags; @@ -807,9 +810,10 @@ static void its_free_tables(struct its_node *its) int i; for (i = 0; i < GITS_BASER_NR_REGS; i++) { - if (its->tables[i]) { - free_page((unsigned long)its->tables[i]); - its->tables[i] = NULL; + if (its->tables[i].base) { + free_pages((unsigned long)its->tables[i].base, + get_order(its->tables[i].size)); + its->tables[i].base = NULL; } } } @@ -880,6 +884,7 @@ retry_alloc_baser: if (alloc_pages > GITS_BASER_PAGES_MAX) { alloc_pages = GITS_BASER_PAGES_MAX; order = get_order(GITS_BASER_PAGES_MAX * psz); + alloc_size = (1 << order) * PAGE_SIZE; pr_warn("%s: Device Table too large, reduce its page order to %u (%u pages)\n", node_name, order, alloc_pages); } @@ -890,7 +895,8 @@ retry_alloc_baser: goto out_free; } - its->tables[i] = base; + its->tables[i].base = base; + its->tables[i].size = alloc_size; retry_baser: val = (virt_to_phys(base) |
The current ITS driver has a memory leak in its_free_tables(). It happens on tear down path of the driver when its_probe() call fails. its_free_tables() should free the exact number of pages that have been allocated, not just a single page as current code does. This patch records the memory size for each ITS_BASERn at the time of page allocation and uses the same size information when freeing pages to fix the issue. Signed-off-by: Shanker Donthineni <shankerd@codeaurora.org> --- [v1]->[v2]: -rebase on tip of https://git.kernel.org/cgit/linux/kernel/git/maz/arm-platforms.git/commit/?h=irq/its-fixes-4.5&id=25ffe003bdc24492204e099fca0a467641d0fe1a -refresh alloc_size value whenever page order changes. drivers/irqchip/irq-gic-v3-its.c | 16 +++++++++++----- 1 file changed, 11 insertions(+), 5 deletions(-)