[PATCH 0/2] Recover from failure to probe GPU
Alex Deucher
alexdeucher at gmail.com
Tue Dec 27 17:04:25 UTC 2022
On Tue, Dec 27, 2022 at 10:40 AM Alex Deucher <alexdeucher at gmail.com> wrote:
>
> On Sun, Dec 25, 2022 at 10:31 AM Christian König
> <christian.koenig at amd.com> wrote:
> >
> > Am 24.12.22 um 10:34 schrieb Thomas Zimmermann:
> > > Hi
> > >
> > > Am 22.12.22 um 19:30 schrieb Mario Limonciello:
> > >> One of the first thing that KMS drivers do during initialization is
> > >> destroy the system firmware framebuffer by means of
> > >> `drm_aperture_remove_conflicting_pci_framebuffers`
> > >>
> > >> This means that if for any reason the GPU failed to probe the user
> > >> will be stuck with at best a screen frozen at the last thing that
> > >> was shown before the KMS driver continued it's probe.
> > >>
> > >> The problem is most pronounced when new GPU support is introduced
> > >> because users will need to have a recent linux-firmware snapshot
> > >> on their system when they boot a kernel with matching support.
> > >>
> > >> However the problem is further exaggerated in the case of amdgpu because
> > >> it has migrated to "IP discovery" where amdgpu will attempt to load
> > >> on "ALL" AMD GPUs even if the driver is missing support for IP blocks
> > >> contained in that GPU.
> > >>
> > >> IP discovery requires some probing and isn't run until after the
> > >> framebuffer has been destroyed.
> > >>
> > >> This means a situation can occur where a user purchases a new GPU not
> > >> yet supported by a distribution and when booting the installer it will
> > >> "freeze" even if the distribution doesn't have the matching kernel
> > >> support
> > >> for those IP blocks.
> > >>
> > >> The perfect example of this is Ubuntu 21.10 and the new dGPUs just
> > >> launched by AMD. The installation media ships with kernel 5.19 (which
> > >> has IP discovery) but the amdgpu support for those IP blocks landed in
> > >> kernel 6.0. The matching linux-firmware was released after 21.10's
> > >> launch.
> > >> The screen will freeze without nomodeset. Even if a user manages to
> > >> install
> > >> and then upgrades to kernel 6.0 after install they'll still have the
> > >> problem of missing firmware, and the same experience.
> > >>
> > >> This is quite jarring for users, particularly if they don't know
> > >> that they have to use "nomodeset" to install.
> > >>
> > >> To help the situation, allow drivers to re-run the init process for the
> > >> firmware framebuffer during a failed probe. As this problem is most
> > >> pronounced with amdgpu, this is the only driver changed.
> > >>
> > >> But if this makes sense more generally for other KMS drivers, the call
> > >> can be added to the cleanup routine for those too.
> > >
> > > Just a quick drive-by comment: as Javier noted, at some point while
> > > probing, your driver has changed the device' state and the system FB
> > > will be gone. you cannot reestablish the sysfb after that.
> >
> > I was about to note exactly that as well. This effort here is
> > unfortunately pretty pointless.
> >
> > >
> > > You are, however free to read device state at any time, as long as it
> > > has no side effects.
> > >
> > > So why not just move the call to
> > > drm_aperture_remove_conflicting_pci_framebuffers() to a later point
> > > when you know that your driver supports the hardware? That's the
> > > solution we always proposed to this kind of problem. It's safe and
> > > won't require any changes to the aperture helpers.
> >
> > if I'm not completely mistaken that's a little bit tricky. Currently
> > it's not possible to read the discovery table before disabling the VGA
> > and/or current framebuffer.
> >
> > We might be able to do this, but it's probably not easy.
>
>
> It should be possible. It's populated by the PSP/VBIOS at power up,
> so all you need to do is read the right offset in vram. For
> firmwares, we currently read them from the filesystem from the
> relevant IP code, but we could also just read it in amdgpu_discovery.c
> when we walk the IP discovery table.
I think something like this would do the trick:
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 2017b3466612..45aee27ab6b1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -2141,6 +2141,11 @@ static int amdgpu_device_ip_early_init(struct
amdgpu_device *adev)
break;
}
+ /* Get rid of things like offb */
+ r = drm_aperture_remove_conflicting_pci_framebuffers(pdev,
&amdgpu_kms_driver);
+ if (r)
+ return r;
+
if (amdgpu_has_atpx() &&
(amdgpu_is_atpx_hybrid() ||
amdgpu_has_atpx_dgpu_power_cntl()) &&
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index b8cfa48fb296..4e74d7abc3c2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -2123,11 +2123,6 @@ static int amdgpu_pci_probe(struct pci_dev *pdev,
}
#endif
- /* Get rid of things like offb */
- ret = drm_aperture_remove_conflicting_pci_framebuffers(pdev,
&amdgpu_kms_driver);
- if (ret)
- return ret;
-
adev = devm_drm_dev_alloc(&pdev->dev, &amdgpu_kms_driver,
typeof(*adev), ddev);
if (IS_ERR(adev))
return PTR_ERR(adev);
>
> Alex
>
>
> >
> > Regards,
> > Christian.
> >
> >
> > >
> > > Best regards
> > > Thomas
> > >
> > >>
> > >> Here is a sample of what happens with missing GPU firmware and this
> > >> series:
> > >>
> > >> [ 5.950056] amdgpu 0000:63:00.0: vgaarb: deactivate vga console
> > >> [ 5.950114] amdgpu 0000:63:00.0: enabling device (0006 -> 0007)
> > >> [ 5.950883] [drm] initializing kernel modesetting (YELLOW_CARP
> > >> 0x1002:0x1681 0x17AA:0x22F1 0xD2).
> > >> [ 5.952954] [drm] register mmio base: 0xB0A00000
> > >> [ 5.952958] [drm] register mmio size: 524288
> > >> [ 5.954633] [drm] add ip block number 0 <nv_common>
> > >> [ 5.954636] [drm] add ip block number 1 <gmc_v10_0>
> > >> [ 5.954637] [drm] add ip block number 2 <navi10_ih>
> > >> [ 5.954638] [drm] add ip block number 3 <psp>
> > >> [ 5.954639] [drm] add ip block number 4 <smu>
> > >> [ 5.954641] [drm] add ip block number 5 <dm>
> > >> [ 5.954642] [drm] add ip block number 6 <gfx_v10_0>
> > >> [ 5.954643] [drm] add ip block number 7 <sdma_v5_2>
> > >> [ 5.954644] [drm] add ip block number 8 <vcn_v3_0>
> > >> [ 5.954645] [drm] add ip block number 9 <jpeg_v3_0>
> > >> [ 5.954663] amdgpu 0000:63:00.0: amdgpu: Fetched VBIOS from VFCT
> > >> [ 5.954666] amdgpu: ATOM BIOS: 113-REMBRANDT-X37
> > >> [ 5.954677] [drm] VCN(0) decode is enabled in VM mode
> > >> [ 5.954678] [drm] VCN(0) encode is enabled in VM mode
> > >> [ 5.954680] [drm] JPEG decode is enabled in VM mode
> > >> [ 5.954681] amdgpu 0000:63:00.0: amdgpu: Trusted Memory Zone (TMZ)
> > >> feature disabled as experimental (default)
> > >> [ 5.954683] amdgpu 0000:63:00.0: amdgpu: PCIE atomic ops is not
> > >> supported
> > >> [ 5.954724] [drm] vm size is 262144 GB, 4 levels, block size is
> > >> 9-bit, fragment size is 9-bit
> > >> [ 5.954732] amdgpu 0000:63:00.0: amdgpu: VRAM: 512M
> > >> 0x000000F400000000 - 0x000000F41FFFFFFF (512M used)
> > >> [ 5.954735] amdgpu 0000:63:00.0: amdgpu: GART: 1024M
> > >> 0x0000000000000000 - 0x000000003FFFFFFF
> > >> [ 5.954738] amdgpu 0000:63:00.0: amdgpu: AGP: 267419648M
> > >> 0x000000F800000000 - 0x0000FFFFFFFFFFFF
> > >> [ 5.954747] [drm] Detected VRAM RAM=512M, BAR=512M
> > >> [ 5.954750] [drm] RAM width 256bits LPDDR5
> > >> [ 5.954834] [drm] amdgpu: 512M of VRAM memory ready
> > >> [ 5.954838] [drm] amdgpu: 15680M of GTT memory ready.
> > >> [ 5.954873] [drm] GART: num cpu pages 262144, num gpu pages 262144
> > >> [ 5.955333] [drm] PCIE GART of 1024M enabled (table at
> > >> 0x000000F41FC00000).
> > >> [ 5.955502] amdgpu 0000:63:00.0: Direct firmware load for
> > >> amdgpu/yellow_carp_toc.bin failed with error -2
> > >> [ 5.955505] amdgpu 0000:63:00.0: amdgpu: fail to request/validate
> > >> toc microcode
> > >> [ 5.955510] [drm:psp_sw_init [amdgpu]] *ERROR* Failed to load psp
> > >> firmware!
> > >> [ 5.955725] [drm:amdgpu_device_init.cold [amdgpu]] *ERROR* sw_init
> > >> of IP block <psp> failed -2
> > >> [ 5.955952] amdgpu 0000:63:00.0: amdgpu: amdgpu_device_ip_init failed
> > >> [ 5.955954] amdgpu 0000:63:00.0: amdgpu: Fatal error during GPU init
> > >> [ 5.955957] amdgpu 0000:63:00.0: amdgpu: amdgpu: finishing device.
> > >> [ 5.971162] efifb: probing for efifb
> > >> [ 5.971281] efifb: showing boot graphics
> > >> [ 5.974803] efifb: framebuffer at 0x910000000, using 20252k, total
> > >> 20250k
> > >> [ 5.974805] efifb: mode is 2880x1800x32, linelength=11520, pages=1
> > >> [ 5.974807] efifb: scrolling: redraw
> > >> [ 5.974807] efifb: Truecolor: size=8:8:8:8, shift=24:16:8:0
> > >> [ 5.974974] Console: switching to colour frame buffer device 180x56
> > >> [ 5.978181] fb0: EFI VGA frame buffer device
> > >> [ 5.978199] amdgpu: probe of 0000:63:00.0 failed with error -2
> > >> [ 5.978285] [drm] amdgpu: ttm finalized
> > >>
> > >> Now if the user loads the firmware into the system they can re-load the
> > >> driver or re-attach using sysfs and it gracefully recovers.
> > >>
> > >> [ 665.080480] [drm] Initialized amdgpu 3.49.0 20150101 for
> > >> 0000:63:00.0 on minor 0
> > >> [ 665.090075] fbcon: amdgpudrmfb (fb0) is primary device
> > >> [ 665.090248] [drm] DSC precompute is not needed.
> > >>
> > >> Mario Limonciello (2):
> > >> firmware: sysfb: Allow re-creating system framebuffer after init
> > >> drm/amd: Re-create firmware framebuffer on failure to probe
> > >>
> > >> drivers/firmware/efi/sysfb_efi.c | 6 +++---
> > >> drivers/firmware/sysfb.c | 15 ++++++++++++++-
> > >> drivers/firmware/sysfb_simplefb.c | 4 ++--
> > >> drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c | 2 ++
> > >> include/linux/sysfb.h | 5 +++++
> > >> 5 files changed, 26 insertions(+), 6 deletions(-)
> > >>
> > >>
> > >> base-commit: 830b3c68c1fb1e9176028d02ef86f3cf76aa2476
> > >
> >
More information about the dri-devel
mailing list