[Mesa-stable] [Mesa-dev] [PATCH] clover: Check the return value of pipe_loader_probe() when probing for devices

Samuel Pitoiset samuel.pitoiset at gmail.com
Mon Nov 30 01:10:56 PST 2015


Hi Hans,

I think your mesa is not up-to-date.

Emil pushed a fix for that one 
http://cgit.freedesktop.org/mesa/mesa/commit/?id=f6235171882d18d0b4d11a3d564c6aa3b22af9d1

Maybe this will help for your issue, but surprisingly, it doesn't fix mine.

On 11/30/2015 09:29 AM, Hans de Goede wrote:
> Hi,
>
> On 29-11-15 17:14, Samuel Pitoiset wrote:
>> This patch fixes the issue for me (eg.
>> https://bugs.freedesktop.org/show_bug.cgi?id=93091#c8).
>>
>> Thanks Tom.
>>
>> Tested-by: Samuel Pitoiset <samuel.pitoiset at gmail.com>
>
> I'm seeing a similar crash, but for me the problem is caused by
> pipe_loader_sw_probe returning 1, but not filling in the devs
> pointer. In which case this fix does not help.
>
> I've been using the following fix locally (I've not gotten
> around to submitting this upstream yet, sorry about that):
>
> --- a/src/gallium/auxiliary/pipe-loader/pipe_loader_sw.c
> +++ b/src/gallium/auxiliary/pipe-loader/pipe_loader_sw.c
> @@ -222,7 +222,7 @@ pipe_loader_sw_probe(struct pipe_loader_device
> **devs, int n
>   {
>      int i = 1;
>
> -   if (i < ndev) {
> +   if (ndev) {
>         if (!pipe_loader_sw_probe_null(devs)) {
>            i--;
>         }
>
>
> The problem is that if pipe_loader_sw_probe gets called with
> ndev=1 it will not set devs, but will still return 1, leading
> to ... (continued below)
>
>
>
>
>>
>> On 11/28/2015 03:52 AM, Tom Stellard wrote:
>>> When probing for devices, clover will call pipe_loader_probe() twice.
>>> The first time to retrieve the number of devices, and then second time
>>> to retrieve the device structures.
>>>
>>> We currently assume that the return value of both calls will be the
>>> same, but this will not be the case if a device happens to disappear
>>> between the two calls.
>>>
>>> This patch removes this assumption and checks the return value of the
>>> second pipe_loader_probe() call to ensure it does not try to initialize
>>> devices that no longer exits.
>>>
>>> CC: <mesa-stable at lists.freedesktop.org>
>>> ---
>>>   src/gallium/state_trackers/clover/core/platform.cpp | 5 +++--
>>>   1 file changed, 3 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/src/gallium/state_trackers/clover/core/platform.cpp
>>> b/src/gallium/state_trackers/clover/core/platform.cpp
>>> index 328b71c..689d692 100644
>>> --- a/src/gallium/state_trackers/clover/core/platform.cpp
>>> +++ b/src/gallium/state_trackers/clover/core/platform.cpp
>>> @@ -28,9 +28,10 @@ platform::platform() : adaptor_range(evals(), devs) {
>>>      int n = pipe_loader_probe(NULL, 0);
>>>      std::vector<pipe_loader_device *> ldevs(n);
>>>
>>> -   pipe_loader_probe(&ldevs.front(), n);
>>> +   n = pipe_loader_probe(&ldevs.front(), n);
>>>
>>> -   for (pipe_loader_device *ldev : ldevs) {
>>> +   for (int i = 0; i < n; ++i) {
>>> +      pipe_loader_device *ldev = ldevs[i];
>>>         try {
>>>            devs.push_back(create<device>(*this, ldev));
>
> create getting called with a ldev value of NULL here.
>
> Regards,
>
> Hans
>
>
>>>         } catch (error &) {
>>>
>> _______________________________________________
>> mesa-dev mailing list
>> mesa-dev at lists.freedesktop.org
>> http://lists.freedesktop.org/mailman/listinfo/mesa-dev

-- 
-Samuel


More information about the mesa-stable mailing list