linux-kernel - Re: [PATCH V1] accel/amdxdna: Fix cu_idx being cleared by memset() during command setup

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <98299eea-9da5-f887-521e-402a735758be@amd.com>
Date: Wed, 10 Dec 2025 09:24:30 -0800
From: Lizhi Hou <lizhi.hou@....com>
To: "Mario Limonciello (AMD) (kernel.org)" <superm1@...nel.org>,
	<ogabbay@...nel.org>, <quic_jhugo@...cinc.com>,
	<dri-devel@...ts.freedesktop.org>, <maciej.falkowski@...ux.intel.com>
CC: <linux-kernel@...r.kernel.org>, <max.zhen@....com>, <sonal.santan@....com>
Subject: Re: [PATCH V1] accel/amdxdna: Fix cu_idx being cleared by memset()
 during command setup

Applied to drm-misc-next.

On 12/9/25 21:37, Mario Limonciello (AMD) (kernel.org) wrote:
>
>
> On 12/9/2025 3:16 PM, Lizhi Hou wrote:
>> For one command type, cu_idx is assigned before calling memset() on the
>> command structure. This results in cu_idx being overwritten, causing the
>> firmware to receive an incomplete or invalid command and leading to
>> unexpected command failures.
>>
>> Fix this by moving the memset() call before initializing cu_idx so that
>> all fields are populated in the correct order.
>>
>> Fixes: 71829d7f2f70 ("accel/amdxdna: Use MSG_OP_CHAIN_EXEC_NPU when 
>> supported")
>> Signed-off-by: Lizhi Hou <lizhi.hou@....com>
> Reviewed-by: Mario Limonciello (AMD) <superm1@...nel.org>
>> ---
>>   drivers/accel/amdxdna/aie2_message.c | 8 ++++----
>>   1 file changed, 4 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/accel/amdxdna/aie2_message.c 
>> b/drivers/accel/amdxdna/aie2_message.c
>> index a75156800467..03b75757a6e6 100644
>> --- a/drivers/accel/amdxdna/aie2_message.c
>> +++ b/drivers/accel/amdxdna/aie2_message.c
>> @@ -652,6 +652,7 @@ aie2_cmdlist_fill_npu_cf(struct amdxdna_gem_obj 
>> *cmd_bo, void *slot, size_t *siz
>>       u32 cmd_len;
>>       void *cmd;
>>   +    memset(npu_slot, 0, sizeof(*npu_slot));
>>       cmd = amdxdna_cmd_get_payload(cmd_bo, &cmd_len);
>>       if (*size < sizeof(*npu_slot) + cmd_len)
>>           return -EINVAL;
>> @@ -660,7 +661,6 @@ aie2_cmdlist_fill_npu_cf(struct amdxdna_gem_obj 
>> *cmd_bo, void *slot, size_t *siz
>>       if (npu_slot->cu_idx == INVALID_CU_IDX)
>>           return -EINVAL;
>>   -    memset(npu_slot, 0, sizeof(*npu_slot));
>>       npu_slot->type = EXEC_NPU_TYPE_NON_ELF;
>>       npu_slot->arg_cnt = cmd_len / sizeof(u32);
>>       memcpy(npu_slot->args, cmd, cmd_len);
>> @@ -677,6 +677,7 @@ aie2_cmdlist_fill_npu_dpu(struct amdxdna_gem_obj 
>> *cmd_bo, void *slot, size_t *si
>>       u32 cmd_len;
>>       u32 arg_sz;
>>   +    memset(npu_slot, 0, sizeof(*npu_slot));
>>       sn = amdxdna_cmd_get_payload(cmd_bo, &cmd_len);
>>       arg_sz = cmd_len - sizeof(*sn);
>>       if (cmd_len < sizeof(*sn) || arg_sz > MAX_NPU_ARGS_SIZE)
>> @@ -689,7 +690,6 @@ aie2_cmdlist_fill_npu_dpu(struct amdxdna_gem_obj 
>> *cmd_bo, void *slot, size_t *si
>>       if (npu_slot->cu_idx == INVALID_CU_IDX)
>>           return -EINVAL;
>>   -    memset(npu_slot, 0, sizeof(*npu_slot));
>>       npu_slot->type = EXEC_NPU_TYPE_PARTIAL_ELF;
>>       npu_slot->inst_buf_addr = sn->buffer;
>>       npu_slot->inst_size = sn->buffer_size;
>> @@ -709,6 +709,7 @@ aie2_cmdlist_fill_npu_preempt(struct 
>> amdxdna_gem_obj *cmd_bo, void *slot, size_t
>>       u32 cmd_len;
>>       u32 arg_sz;
>>   +    memset(npu_slot, 0, sizeof(*npu_slot));
>>       pd = amdxdna_cmd_get_payload(cmd_bo, &cmd_len);
>>       arg_sz = cmd_len - sizeof(*pd);
>>       if (cmd_len < sizeof(*pd) || arg_sz > MAX_NPU_ARGS_SIZE)
>> @@ -721,7 +722,6 @@ aie2_cmdlist_fill_npu_preempt(struct 
>> amdxdna_gem_obj *cmd_bo, void *slot, size_t
>>       if (npu_slot->cu_idx == INVALID_CU_IDX)
>>           return -EINVAL;
>>   -    memset(npu_slot, 0, sizeof(*npu_slot));
>>       npu_slot->type = EXEC_NPU_TYPE_PREEMPT;
>>       npu_slot->inst_buf_addr = pd->inst_buf;
>>       npu_slot->save_buf_addr = pd->save_buf;
>> @@ -745,6 +745,7 @@ aie2_cmdlist_fill_npu_elf(struct amdxdna_gem_obj 
>> *cmd_bo, void *slot, size_t *si
>>       u32 cmd_len;
>>       u32 arg_sz;
>>   +    memset(npu_slot, 0, sizeof(*npu_slot));
>>       pd = amdxdna_cmd_get_payload(cmd_bo, &cmd_len);
>>       arg_sz = cmd_len - sizeof(*pd);
>>       if (cmd_len < sizeof(*pd) || arg_sz > MAX_NPU_ARGS_SIZE)
>> @@ -753,7 +754,6 @@ aie2_cmdlist_fill_npu_elf(struct amdxdna_gem_obj 
>> *cmd_bo, void *slot, size_t *si
>>       if (*size < sizeof(*npu_slot) + arg_sz)
>>           return -EINVAL;
>>   -    memset(npu_slot, 0, sizeof(*npu_slot));
>>       npu_slot->type = EXEC_NPU_TYPE_ELF;
>>       npu_slot->inst_buf_addr = pd->inst_buf;
>>       npu_slot->save_buf_addr = pd->save_buf;
>