[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <32eb018069a079fd82e359d0ed4e1b1e@codeaurora.org>
Date: Mon, 16 Nov 2020 17:33:34 +0800
From: Wen Gong <wgong@...eaurora.org>
To: Kalle Valo <kvalo@...eaurora.org>
Cc: kernel test robot <rong.a.chen@...el.com>, kbuild-all@...ts.01.org,
linux-kernel@...r.kernel.org, ath10k@...ts.infradead.org
Subject: Re: drivers/net/wireless/ath/ath10k/sdio.c:2234:2: warning:
Non-boolean value returned from function returning bool
On 2020-11-16 15:53, Kalle Valo wrote:
> + ath10k list
>
> kernel test robot <rong.a.chen@...el.com> writes:
>
>> tree:
>> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
>> master
>> head: f01c30de86f1047e9bae1b1b1417b0ce8dcd15b1
>> commit: 3c45f21af84eb05a355919abc80cf70a3a681cee ath10k: sdio: add
>> firmware coredump support
>> compiler: nios2-linux-gcc (GCC) 9.3.0
>>
>> If you fix the issue, kindly add following tag as appropriate
>> Reported-by: kernel test robot <rong.a.chen@...el.com>
>>
>>
>> cppcheck possible warnings: (new ones prefixed by >>, may not real
>> problems)
>>
>>>> drivers/net/wireless/ath/ath10k/sdio.c:2234:2: warning: Non-boolean
>> value returned from function returning bool
>> [returnNonBoolInBooleanFunction]
>> return param & HI_OPTION_SDIO_CRASH_DUMP_ENHANCEMENT_FW;
>> ^
>
> Is this really a problem? I guess we could change that to "!!(param &
> HI_OPTION_SDIO_CRASH_DUMP_ENHANCEMENT_FW)" but how is that better and
> does it make any practical difference when
> ath10k_sdio_is_fast_dump_supported() returns a boolean anyway?
It should not be a real problem by test before.
ath10k_sdio_is_fast_dump_supported return true in test before.
and fast_dump get true from ath10k_sdio_is_fast_dump_supported in
ath10k_sdio_fw_crashed_dump.
the risk maybe like this:
u32 param in ath10k_sdio_is_fast_dump_supported is 4 bytes, and bool is
1 byte.
if Gcc does not do any convertion from u32 to bool, then the bool value
will only
get the lower byte(byte 0) of the 4 bytes, then maybe get an mistake
result.
I did a disassemble of sdio.o with/without below change, and dump
sdio.o, it have no difference in this place:
I change send a patch with below change if needed.
@@ -2231,7 +2231,7 @@ static bool
ath10k_sdio_is_fast_dump_supported(struct ath10k *ar)
ath10k_dbg(ar, ATH10K_DBG_SDIO, "sdio hi_option_flag2 %x\n",
param);
- return param & HI_OPTION_SDIO_CRASH_DUMP_ENHANCEMENT_FW;
+ return !!(param & HI_OPTION_SDIO_CRASH_DUMP_ENHANCEMENT_FW);
}
objdump -Sl sdio.o > sdio.o-Sl.txt
0000000000002b00 <ath10k_sdio_fw_crashed_dump>:
void ath10k_sdio_fw_crashed_dump(struct ath10k *ar)
{
...
/home/wgong/ath11k/ath-master/ath/drivers/net/wireless/ath/ath10k/sdio.c:2234
return param & HI_OPTION_SDIO_CRASH_DUMP_ENHANCEMENT_FW;
2b75: 8b 44 24 44 mov 0x44(%rsp),%eax
2b79: 25 00 08 00 00 and $0x800,%eax //"the SF, ZF, and PF
flags are set according to the result".ZF will be 0 if true.
2b7e: 89 44 24 2c mov %eax,0x2c(%rsp)
2b82: 40 0f 95 c5 setne %bpl //"Set byte if not equal
(ZF=0)".stored the ZF flag to bpl as value of fast_dump.
ath10k_sdio_fw_crashed_dump():
/home/wgong/ath11k/ath-master/ath/drivers/net/wireless/ath/ath10k/sdio.c:2483
char guid[UUID_STRING_LEN + 1];
bool fast_dump;
fast_dump = ath10k_sdio_is_fast_dump_supported(ar);
if (fast_dump)
2b86: 0f 85 c3 03 00 00 jne 2f4f
<ath10k_sdio_fw_crashed_dump+0x44f> //"Jump near if not equal (ZF=0)"
...
/home/wgong/ath11k/ath-master/ath/drivers/net/wireless/ath/ath10k/sdio.c:2499
ath10k_sdio_dump_registers(ar, crash_data, fast_dump);
2be5: 40 0f b6 d5 movzbl %bpl,%edx //use the bpl stored as
value of fast_dump
2be9: 48 89 de mov %rbx,%rsi
2bec: 4c 89 ff mov %r15,%rdi
2bef: e8 8c fb ff ff callq 2780 <ath10k_sdio_dump_registers>
Powered by blists - more mailing lists