Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Jaewon Kim <jaewon31.kim@samsung.com>
To:     jstultz@google.com, tjmercier@google.com, sumit.semwal@linaro.org,
        daniel.vetter@ffwll.ch, akpm@linux-foundation.org,
        hannes@cmpxchg.org, mhocko@kernel.org
Cc:     linux-mm@kvack.org, linux-kernel@vger.kernel.org,
        jaewon31.kim@gmail.com, Jaewon Kim <jaewon31.kim@samsung.com>
Subject: [PATCH v3] dma-buf/heaps: system_heap: avoid too much allocation
Date:   Mon, 10 Apr 2023 16:32:28 +0900
Message-Id: <20230410073228.23043-1-jaewon31.kim@samsung.com>
Content-Type: text/plain; charset="utf-8"
CMS-TYPE: 101P
DLP-Filter: Pass
References: <CGME20230410073304epcas1p4cf3079b096994d69472b7801bd530bc7@epcas1p4.samsung.com>
Precedence: bulk

Normal free:212600kB min:7664kB low:57100kB high:106536kB
  reserved_highatomic:4096KB active_anon:276kB inactive_anon:180kB
  active_file:1200kB inactive_file:0kB unevictable:2932kB
  writepending:0kB present:4109312kB managed:3689488kB mlocked:2932kB
  pagetables:13600kB bounce:0kB free_pcp:0kB local_pcp:0kB
  free_cma:200844kB
Out of memory and no killable processes...
Kernel panic - not syncing: System is deadlocked on memory

An OoM panic was reported. The log shows there were only native
processes which are non-killable as OOM_SCORE_ADJ_MIN. After looking
into the dump, I've found the dma-buf system heap was trying to allocate
a huge size. It seems to be a signed negative value.

dma_heap_ioctl_allocate(inline)
    |  heap_allocation = 0xFFFFFFC02247BD38 -> (
    |    len = 0xFFFFFFFFE7225100,

To avoid this invalid request, check if the requested size is bigger
than system total memory. Actually the old ion system heap had similar
policy with commit c9e8440eca61 ("staging: ion: Fix overflow and list
bugs in system heap").

Even with this sanity check, there is still risk of too much allocations
from the system_heap. Allocating multiple big size buffers may cause
oom. Add __GFP_RETRY_MAYFAIL. With this gfp, the allocation may fail,
but we can avoid oom panic.

Signed-off-by: Jaewon Kim <jaewon31.kim@samsung.com>
Acked-by: John Stultz <jstultz@google.com>
Reviewed-by: T.J. Mercier <tjmercier@google.com>
---
 drivers/dma-buf/heaps/system_heap.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/drivers/dma-buf/heaps/system_heap.c b/drivers/dma-buf/heaps/system_heap.c
index 920db302a273..583da8948679 100644
--- a/drivers/dma-buf/heaps/system_heap.c
+++ b/drivers/dma-buf/heaps/system_heap.c
@@ -41,7 +41,7 @@ struct dma_heap_attachment {
 	bool mapped;
 };
 
-#define LOW_ORDER_GFP (GFP_HIGHUSER | __GFP_ZERO)
+#define LOW_ORDER_GFP (GFP_HIGHUSER | __GFP_ZERO | __GFP_RETRY_MAYFAIL)
 #define HIGH_ORDER_GFP  (((GFP_HIGHUSER | __GFP_ZERO | __GFP_NOWARN \
 				| __GFP_NORETRY) & ~__GFP_RECLAIM) \
 				| __GFP_COMP)
@@ -350,6 +350,9 @@ static struct dma_buf *system_heap_allocate(struct dma_heap *heap,
 	struct page *page, *tmp_page;
 	int i, ret = -ENOMEM;
 
+	if (len / PAGE_SIZE > totalram_pages())
+		return ERR_PTR(-ENOMEM);
+
 	buffer = kzalloc(sizeof(*buffer), GFP_KERNEL);
 	if (!buffer)
 		return ERR_PTR(-ENOMEM);
-- 
2.17.1