Re: [Mesa-dev] Question about display_gpu and render gpu in dri3_create_screen

Mon, 13 Sep 2021 23:35:03 -0700 

Mhm, that looks like it's something I should look into.

How exactly are you triggering this and which kernel version are you using?

Thanks,
Christian.

Am 14.09.21 um 03:47 schrieb Luc Ma:

Yes, I did. It crashed the kernel as follows
Sep 14 09:12:34 kernel: [  506.676925][ 4] [ T1045] Hardware name: WEIBU F20A8/F20A8, BIOS 0.1.3 2021-07-06_20:59:09 Sep 14 09:12:34 kernel: [  506.685091][ 4] [ T1045] pstate: 80000005 (Nzcv daif -PAN -UAO) 
Sep 14 09:12:34 kernel: [  506.691001][ 4] [ T1045] pc : 0x0
Sep 14 09:12:34 kernel: [  506.694341][ 4] [ T1045] lr : drm_gem_map_dma_buf+0xc0/0x118 [drm] 
Sep 14 09:12:34 kernel: [  506.700509][ 4] [ T1045] sp : ffffffa0f4993b10
Sep 14 09:12:34 kernel: [  506.704941][ 4] [ T1045] x29: ffffffa0f4993b10 x28: ffffffa0f4993cd8 Sep 14 09:12:34 kernel: [  506.711370][ 4] [ T1045] x27: 000000000000000c x26: ffffffa0f33e5200 Sep 14 09:12:34 kernel: [  506.717800][ 4] [ T1045] x25: ffffffa0e7452630 x24: ffffffa0f5a9f000 Sep 14 09:12:34 kernel: [  506.724229][ 4] [ T1045] x23: ffffffa0f33e5328 x22: ffffffa0e820f180 Sep 14 09:12:34 kernel: [  506.730658][ 4] [ T1045] x21: ffffffa0e820f180 x20: 0000000000000000 Sep 14 09:12:34 kernel: [  506.737087][ 4] [ T1045] x19: ffffffa0e820f180 x18: 0000000000000000 Sep 14 09:12:34 kernel: [  506.743516][ 4] [ T1045] x17: 0000000000000000 x16: 0000000000000000 Sep 14 09:12:34 kernel: [  506.749945][ 4] [ T1045] x15: 0000000000000000 x14: 0000000000000000 Sep 14 09:12:34 kernel: [  506.756373][ 4] [ T1045] x13: 0000000000000000 x12: 0000000000000000 Sep 14 09:12:34 kernel: [  506.762802][ 4] [ T1045] x11: 0000000000000000 x10: ffffffa0f43d1938 Sep 14 09:12:34 kernel: [  506.769231][ 4] [ T1045] x9 : 0000000000000000 x8 : ffffffa0e820f200 Sep 14 09:12:34 kernel: [  506.775659][ 4] [ T1045] x7 : 0000000000000000 x6 : 000000000000003f Sep 14 09:12:34 kernel: [  506.782088][ 4] [ T1045] x5 : 0000000000000040 x4 : 0000000000000000 Sep 14 09:12:34 kernel: [  506.788517][ 4] [ T1045] x3 : 0000000000000000 x2 : ffffffc008d553b0 Sep 14 09:12:34 kernel: [  506.794945][ 4] [ T1045] x1 : 0000000000000000 x0 : ffffffa0f311b000 
Sep 14 09:12:34 kernel: [  506.801375][ 4] [ T1045] Call trace:
Sep 14 09:12:34 kernel: [  506.804938][ 4] [ T1045]  0x0
Sep 14 09:12:34 kernel: [  506.807899][ 4] [ T1045]  dma_buf_map_attachment+0x60/0xb0 Sep 14 09:12:34 kernel: [  506.813398][ 4] [ T1045]  drm_gem_prime_import_dev+0x7c/0x138 [drm] Sep 14 09:12:34 kernel: [  506.819675][ 4] [ T1045]  drm_gem_prime_fd_to_handle+0x1b4/0x1d8 [drm] Sep 14 09:12:34 kernel: [  506.826213][ 4] [ T1045]  drm_prime_fd_to_handle_ioctl+0x24/0x38 [drm] Sep 14 09:12:34 kernel: [  506.832750][ 4] [ T1045]  drm_ioctl_kernel+0x84/0xd8 [drm] Sep 14 09:12:34 kernel: [  506.838245][ 4] [ T1045]  drm_ioctl+0x218/0x408 [drm] Sep 14 09:12:34 kernel: [  506.843326][ 4] [ T1045]  radeon_drm_ioctl+0x50/0x88 [radeon] Sep 14 09:12:34 kernel: [  506.849063][ 4] [ T1045]  do_vfs_ioctl+0x394/0x7e8 
Sep 14 09:12:34 kernel: [  506.853843][ 4] [ T1045]  ksys_ioctl+0x78/0xa8
Sep 14 09:12:34 kernel: [  506.858276][ 4] [ T1045]  __arm64_sys_ioctl+0x1c/0x28 Sep 14 09:12:34 kernel: [  506.863318][ 4] [ T1045]  el0_svc_common.constprop.0+0x68/0x168 Sep 14 09:12:34 kernel: [  506.869226][ 4] [ T1045]  el0_svc_handler+0x8c/0x98 
Sep 14 09:12:34 kernel: [  506.874093][ 4] [ T1045]  el0_svc+0x8/0xc
Sep 14 09:12:34 kernel: [  506.878093][ 4] [ T1045] Code: bad PC value
Sep 14 09:12:34 kernel: [  506.882265][ 4] [ T1045] ---[ end trace 3888e65eac0454cc ]---
I'm not familiar with the kernel, but I suspect it may be a kmd problem. It seems like something is missing.
On Mon, 13 Sept 2021 at 20:57, Michel Dänzer <mic...@daenzer.net <mailto:mic...@daenzer.net>> wrote: 

    On 2021-09-13 14:40, Luc Ma wrote:
    > Hello,
    >
    > I recently tried multi-GPU support on the mesa gallium
    drivers(glx=dri). When I exported the env `DRI_PRIME=1`, I found
    that it didn't work with two different drivers loaded at the same
    time.
    > because there are different driver names in my case. display_gpu
    driver name is "r600" while render_gpu driver name is "nouveau".
    It failed to create display gpu screen
    >
    > if (strcmp(driverName, driverNameDisplayGPU) == 0) {
    >             psc->driScreenDisplayGPU =
    >  psc->image_driver->createNewScreen2(screen, psc->fd_display_gpu,
    >  pdp->loader_extensions,
    >  extensions,
    >  &driver_configs, psc);
    > }
    >
    > so I am wondering
    >
    > - is it possible to use two GPUs from different vendors on a
    system with gallium? one is for display, another for rendering
    > - is it possible to use two GPUs driven by a shared driver(say
    both "r600") on a system?

    Both should work.


    Did you hit a problem other than psc->driScreenDisplayGPU related
    code being skipped (as is expected ATM with different drivers)?
-- Earthling Michel Dänzer               | https://redhat.com 
    <https://redhat.com>
    Libre software enthusiast             |             Mesa and X
    developer

Reply via email to