Re: [PATCH i-g-t v1] lib/igt_debugfs: Add "errno" to debug print

[Ngai-Mint Kwan <ngai-mint.kwan@linux.intel.com>]

Hi Kamil,

On 2026-07-13 07:45, Kamil Konieczny wrote:
> Hi Ngai-Mint,
> On 2026-07-08 at 15:52:12 -0700, Ngai-Mint Kwan wrote:
>> When igt_debugfs_path() fails under igt_debugfs_dir(), add "errno" to the
>> debug print for additional context of the failure.
>>
>> Signed-off-by: Ngai-Mint Kwan <ngai-mint.kwan@linux.intel.com>
>> ---
>>   lib/igt_debugfs.c | 6 +++++-
>>   1 file changed, 5 insertions(+), 1 deletion(-)
>>
>> diff --git a/lib/igt_debugfs.c b/lib/igt_debugfs.c
>> index ed0a20773..a8c45945c 100644
>> --- a/lib/igt_debugfs.c
>> +++ b/lib/igt_debugfs.c
>> @@ -187,8 +187,12 @@ int igt_debugfs_dir(int device)
>>   	int debugfs_dir_fd;
>>   	char path[200];
>>   
>> -	if (igt_debug_on(!igt_debugfs_path(device, path, sizeof(path))))
>> +	errno = 0;
>> +	if (igt_debug_on(!igt_debugfs_path(device, path, sizeof(path)))) {
>> +		igt_debug("Last errno: %i, %s\n", errno, strerror(errno));
> Problem with this is that errno should be used in place where
> the error occured, as any system function call after it occur
> can change it. imho you should add more debug prints to
> igt_debugfs_path() with errno printing there.
Agreed, but the goal of this patch is narrowly focused to improve 
debugging for
igt_debugfs_dir(). The idea is to provide a hint of the debugfs failure 
so that the
person debugging would have additional context. This also includes the 
possibility
that errno is zero and that the failure is potentially unrelated to 
system function
calls.

Regards,
Ngai-Mint Kwan
> Regards,
> Kamil
>
>> +
>>   		return -1;
>> +	}
>>   
>>   	debugfs_dir_fd = open(path, O_RDONLY);
>>   	igt_debug_on_f(debugfs_dir_fd < 0, "path: %s\n", path);
>> -- 
>> 2.43.0
>>