From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Message-ID: <6a9b5529-6b41-a29b-3608-81be00934fbf@linux.intel.com> Date: Thu, 28 Sep 2023 09:16:23 +0100 MIME-Version: 1.0 Content-Language: en-US To: Umesh Nerlige Ramappa References: <20230922134437.234888-1-tvrtko.ursulin@linux.intel.com> <20230922134437.234888-4-tvrtko.ursulin@linux.intel.com> From: Tvrtko Ursulin In-Reply-To: Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit Subject: Re: [igt-dev] [PATCH i-g-t 03/12] tools/intel_gpu_top: Restore user friendly error message List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: igt-dev@lists.freedesktop.org, Intel-gfx@lists.freedesktop.org, Tvrtko Ursulin Errors-To: igt-dev-bounces@lists.freedesktop.org Sender: "igt-dev" List-ID: On 27/09/2023 21:13, Umesh Nerlige Ramappa wrote: > On Fri, Sep 22, 2023 at 02:44:28PM +0100, Tvrtko Ursulin wrote: >> From: Tvrtko Ursulin >> >> We have a nice error message displayed when an user with insufficient >> permissions tries to run the tool, but that got lost while Meteorlake >> support was added. Bring it back in. >> >> Signed-off-by: Tvrtko Ursulin >> Cc: Umesh Nerlige Ramappa >> --- >> tools/intel_gpu_top.c | 10 +++++++--- >> 1 file changed, 7 insertions(+), 3 deletions(-) >> >> diff --git a/tools/intel_gpu_top.c b/tools/intel_gpu_top.c >> index 87e9681e53b4..e01355f90458 100644 >> --- a/tools/intel_gpu_top.c >> +++ b/tools/intel_gpu_top.c >> @@ -554,9 +554,11 @@ static int get_num_gts(uint64_t type) >> >>         close(fd); >>     } >> -    assert(!errno || errno == ENOENT); >> -    assert(cnt > 0); >> -    errno = 0; >> + >> +    if (!cnt) >> +        cnt = errno; >> +    else >> +        errno = 0; > > ENOENT is the only way this logic is checking for num_gts. > > In this case error is propagated only if cnt == 0. What if cnt=1 and we > get an error (other than ENOENT)? Should we ignore that? If cnt >= 1 then at least one tile was found so the errno happened while probing for further tiles. So on single tile parts it can be ignored. On multi-tile parts it cannot really happen, or even if it happens situation would simply be "why is only one tile showing". If we want to cover this impossible/unlikely case then maybe like this: if (!cnt || (errno && errno != ENOENT)) cnt = -errno; > I had something like this in mind for the regression (and sorry this > fell through the cracks) > > https://patchwork.freedesktop.org/patch/541406/?series=118973&rev=1 Oh back in June! I think yours work too, in which case it's a matter of a style choice with which one to go. I don't have a strong preference - above would be a bit more compact, while I think it still succinctly expresses the failure condition ("nothing found or unexpected error while probing for remote tiles"). Regards, Tvrtko > > Regards, > Umesh > >> >>     return cnt; >> } >> @@ -590,6 +592,8 @@ static int pmu_init(struct engines *engines) >>     engines->fd = -1; >>     engines->num_counters = 0; >>     engines->num_gts = get_num_gts(type); >> +    if (engines->num_gts <= 0) >> +        return -1; >> >>     engines->irq.config = I915_PMU_INTERRUPTS; >>     fd = _open_pmu(type, engines->num_counters, &engines->irq, >> engines->fd); >> -- >> 2.39.2 >>