Re: [PATCH libdrm] xf86drm: Parse the separate files to retrieve the vendor/device info

[Mauro Santos <registo.mailling@gmail.com>]

On 08-11-2016 11:06, Emil Velikov wrote:
> On 1 November 2016 at 18:47, Mauro Santos <registo.mailling@gmail.com> wrote:
>> On 01-11-2016 18:13, Emil Velikov wrote:
>>> From: Emil Velikov <emil.velikov@collabora.com>
>>>
>>> Parsing config sysfs file wakes up the device. The latter of which may
>>> be slow and isn't required to begin with.
>>>
>>> Reading through config is/was required since the revision is not
>>> available by other means, although with a kernel patch in the way we can
>>> 'cheat' temporarily.
>>>
>>> That should be fine, since no open-source project has ever used the
>>> value.
>>>
>>> Cc: Michel Dänzer <michel.daenzer@amd.com>
>>> Cc: Mauro Santos <registo.mailling@gmail.com>
>>> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=98502
>>> Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
>>> ---
>>> Mauro can you apply this against libdrm and rebuild it. You do _not_
>>> need to rebuild mesa afterwords.
>>>
>>> Thanks
>>> ---
>>>  xf86drm.c | 50 +++++++++++++++++++++++++++++++++++---------------
>>>  1 file changed, 35 insertions(+), 15 deletions(-)
>>>
>>> diff --git a/xf86drm.c b/xf86drm.c
>>> index 52add5e..5a5100c 100644
>>> --- a/xf86drm.c
>>> +++ b/xf86drm.c
>>> @@ -2950,25 +2950,45 @@ static int drmParsePciDeviceInfo(const char *d_name,
>>>                                   drmPciDeviceInfoPtr device)
>>>  {
>>>  #ifdef __linux__
>>> +#define ARRAY_SIZE(a) (sizeof(a) / sizeof((a)[0]))
>>> +    static const char *attrs[] = {
>>> +      "revision", /* XXX: make sure it's always first, see note below */
>>> +      "vendor",
>>> +      "device",
>>> +      "subsystem_vendor",
>>> +      "subsystem_device",
>>> +    };
>>>      char path[PATH_MAX + 1];
>>> -    unsigned char config[64];
>>> -    int fd, ret;
>>> +    unsigned int data[ARRAY_SIZE(attrs)];
>>> +    FILE *fp;
>>> +    int ret;
>>>
>>> -    snprintf(path, PATH_MAX, "/sys/class/drm/%s/device/config", d_name);
>>> -    fd = open(path, O_RDONLY);
>>> -    if (fd < 0)
>>> -        return -errno;
>>> +    for (unsigned i = 0; i < ARRAY_SIZE(attrs); i++) {
>>> +        snprintf(path, PATH_MAX, "/sys/class/drm/%s/device/%s",
>>> +                 d_name, attrs[i]);
>>> +        fp = fopen(path, "r");
>>> +        if (!fp) {
>>> +            /* Note: First we check the revision, since older kernels
>>> +             * may not have it. Default to zero in such cases. */
>>> +            if (i == 0) {
>>> +                data[i] = 0;
>>> +                continue;
>>> +            }
>>> +            return -errno;
>>> +        }
>>>
>>> -    ret = read(fd, config, sizeof(config));
>>> -    close(fd);
>>> -    if (ret < 0)
>>> -        return -errno;
>>> +        ret = fscanf(fp, "%x", &data[i]);
>>> +        fclose(fp);
>>> +        if (ret != 1)
>>> +            return -errno;
>>> +
>>> +    }
>>>
>>> -    device->vendor_id = config[0] | (config[1] << 8);
>>> -    device->device_id = config[2] | (config[3] << 8);
>>> -    device->revision_id = config[8];
>>> -    device->subvendor_id = config[44] | (config[45] << 8);
>>> -    device->subdevice_id = config[46] | (config[47] << 8);
>>> +    device->revision_id = data[0] & 0xff;
>>> +    device->vendor_id = data[1] & 0xffff;
>>> +    device->device_id = data[2] & 0xffff;
>>> +    device->subvendor_id = data[3] & 0xffff;
>>> +    device->subdevice_id = data[4] & 0xffff;
>>>
>>>      return 0;
>>>  #else
>>>
>>
>> I have applied this against libdrm 2.4.71 and I don't see any delays
>> when starting firefox/chromium/thunderbird/glxgears.
>>
>> There is also no indication in dmesg that the dGPU is being
>> reinitialized when starting the programs where I've detected the problem.
>>
> Thanks Mauro. Can you give this a try alongside the kernel fix [1] ?
> I'd love to get the latter merged soon(ish).
> Independent of the kernel side, I might need to go another way for
> libdrm/mesa so I'll CC you on future patches.
> 
> Your help is greatly appreciated !
> 
> Thanks
> Emil
> 
> [1] http://patchwork.ozlabs.org/patch/689975/
> 

I have applied the patch on top of kernel 4.8.6 and I'm using the libdrm
with the patch you sent me previously.

With this patch things still seem work fine, I don't see any of the
problems I've seen before, but I don't know how to confirm that the
value from sysfs is now being used by libdrm instead of defaulting to zero.

The values in /sys/class/drm/card{0,1}/device/revision match what is
reported by lspci for the iGPU and dGPU.
I don't know if it's related but the revision file in
/sys/class/pci_bus/0000\:03/device/ does show 0xf1 for both GPUs, which
is different from what is shown in /sys/class/drm/card{0,1}/device/revision.

-- 
Mauro Santos
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel