Re: [PATCH 5.10 0/2] backport i915 fixes to 5.10

[Henning Schild <henning.schild@siemens.com>]

Am Fri, 2 Jun 2023 09:14:08 -0700
schrieb Matt Roper <matthew.d.roper@intel.com>:

> On Fri, Jun 02, 2023 at 06:05:05PM +0200, Henning Schild wrote:
> > This fixes the following problem which was seen on a Tigerlake
> > running Debian 10 with a 5.10 kernel.
> > 
> > [  111.408631] Missing case (val == 65535)
> > [  111.408698] WARNING: CPU: 2 PID: 446 at
> > drivers/gpu/drm/i915/intel_dram.c:95
> > skl_dram_get_dimm_info+0x72/0x1a0 [i915] [  111.408699] Modules
> > linked in: intel_powerclamp coretemp i915(+) joydev kvm_intel kvm
> > hid_generic irqbypass crc32_pclmul snd_hda_intel
> > ghash_clmulni_intel snd_intel_dspcfg snd_hda_codec aesni_intel
> > glue_helper crypto_simd cryptd snd_hwdep snd_hda_core
> > drm_kms_helper uas snd_pcm usb_storage intel_cstate mei_wdt
> > mei_hdcp snd_timer scsi_mod usbhid hid cec wdat_wdt snd mei_me
> > evdev mei intel_uncore rc_core watchdog pcspkr soundcore video
> > acpi_pad acpi_tad button drm fuse configfs loop(+) efi_pstore
> > efivarfs ip_tables x_tables autofs4 ext4 crc32c_generic crc16
> > mbcache jbd2 dm_mod xhci_pci xhci_hcd marvell dwmac_intel stmmac
> > igb e1000e usbcore nvme nvme_core pcs_xpcs phylink libphy
> > i2c_algo_bit t10_pi dca crc_t10dif crct10dif_generic intel_lpss_pci
> > ptp vmd intel_lpss i2c_i801 crct10dif_pclmul pps_core
> > crct10dif_common idma64 usb_common crc32c_intel i2c_smbus [
> > 111.408755] CPU: 2 PID: 446 Comm: (udev-worker) Not tainted
> > 5.10.180 #2 [  111.408756] Hardware name: SIEMENS AG SIMATIC
> > IPC427G/no information, BIOS T29.01.02.D3.0 10/11/2022 [
> > 111.408797] RIP: 0010:skl_dram_get_dimm_info+0x72/0x1a0 [i915] [
> > 111.408799] Code: 01 00 00 0f 84 31 01 00 00 66 3d 00 01 0f 84 27
> > 01 00 00 41 0f b7 d0 48 c7 c6 ba 81 7c c1 48 c7 c7 be 81 7c c1 e8
> > d2 75 89 fa <0f> 0b c6 45 01 00 44 0f b6 4d 00 31 f6 b9 01 00 00 00
> > c1 fb 09 83 [  111.408801] RSP: 0018:ffffa23a40b53b10 EFLAGS:
> > 00010286 [  111.408802] RAX: 0000000000000000 RBX: 000000000000ffff
> > RCX: 0000000000000027 [  111.408803] RDX: ffff947b278a0908 RSI:
> > 0000000000000001 RDI: ffff947b278a0900 [  111.408804] RBP:
> > ffffa23a40b53b78 R08: 0000000000000000 R09: ffffa23a40b53920 [
> > 111.408805] R10: ffffa23a40b53918 R11: 0000000000000003 R12:
> > 0000000000000000 [  111.408806] R13: 000000000000004c R14:
> > ffff9479b1a00000 R15: ffff9479b1a00000 [  111.408808] FS:
> > 00007ff9626478c0(0000) GS:ffff947b27880000(0000)
> > knlGS:0000000000000000 [  111.408809] CS:  0010 DS: 0000 ES: 0000
> > CR0: 0000000080050033 [  111.408810] CR2: 0000555753cbf15c CR3:
> > 0000000108432002 CR4: 0000000000770ee0 [  111.408810] PKRU:
> > 55555554 [  111.408811] Call Trace: [  111.408854]
> > skl_dram_get_channel_info+0x24/0x160 [i915] [  111.408892]
> > intel_dram_detect+0xef/0x630 [i915] [  111.408931]
> > i915_driver_probe+0xb18/0xc40 [i915] [  111.408969]  ?
> > i915_pci_probe+0x3f/0x160 [i915] [  111.408973]
> > local_pci_probe+0x3b/0x80 [  111.408975]
> > pci_device_probe+0xfc/0x1b0 [  111.408979]
> > really_probe+0x26e/0x460 [  111.408981]
> > driver_probe_device+0xb4/0x100 [  111.408983]
> > device_driver_attach+0xa9/0xb0 [  111.408984]  ?
> > device_driver_attach+0xb0/0xb0 [  111.408985]
> > __driver_attach+0xa1/0x140 [  111.408987]  ?
> > device_driver_attach+0xb0/0xb0 [  111.408989]
> > bus_for_each_dev+0x84/0xd0 [  111.408991]
> > bus_add_driver+0x13e/0x200 [  111.408993]
> > driver_register+0x89/0xe0 [  111.409036]  i915_init+0x60/0x75
> > [i915] [  111.409038]  ? 0xffffffffc18b8000 [  111.409041]
> > do_one_initcall+0x56/0x1f0 [  111.409044]
> > do_init_module+0x4a/0x240 [  111.409047]
> > __do_sys_finit_module+0xaa/0x110 [  111.409050]
> > do_syscall_64+0x30/0x40 [  111.409053]
> > entry_SYSCALL_64_after_hwframe+0x61/0xc6 [  111.409054] RIP:
> > 0033:0x7ff962d534f9 [  111.409056] Code: 08 89 e8 5b 5d c3 66 2e 0f
> > 1f 84 00 00 00 00 00 90 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89
> > c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b
> > 0d d7 08 0d 00 f7 d8 64 89 01 48 [  111.409057] RSP:
> > 002b:00007fff57caf5d8 EFLAGS: 00000246 ORIG_RAX: 0000000000000139 [
> >  111.409058] RAX: ffffffffffffffda RBX: 0000555753cad120 RCX:
> > 00007ff962d534f9 [  111.409059] RDX: 0000000000000000 RSI:
> > 00007ff962ee6efd RDI: 0000000000000010 [  111.409060] RBP:
> > 00007ff962ee6efd R08: 0000000000000000 R09: 0000555753c7a320 [
> > 111.409061] R10: 0000000000000010 R11: 0000000000000246 R12:
> > 0000000000020000 [  111.409062] R13: 0000000000000000 R14:
> > 0000555753ca4f30 R15: 0000555752d40e4f [  111.409064] ---[ end
> > trace 2da6ec0bd6f7c3a1 ]---
> > 
> > 
> > José Roberto de Souza (1):
> >   drm/i915/gen11+: Only load DRAM information from pcode
> > 
> > Matt Roper (1):
> >   drm/i915/dg1: Wait for pcode/uncore handshake at startup  
> 
> This second patch should only be needed for discrete GPUs (none of
> which are fully enabled on a 5.10 kernel).  Are you seeing it cause a
> change in behavior on an integrated TGL platform?

When i started looking at things i thought it would be a bisect job,
because Debian 12 with 6.1 is not affected.
Then i found a ubuntu bug and basically just applied the patch that bug
was talking about (p2)

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1933274

which needs p1 for that new function

Having both applied the problem i did see is indeed gone. Not that i
understand what that code does in depth.

I think it is Tigerlake but i always have a hard time with the "lakes"

CPU:
Intel(R) Xeon(R) W-11865MLE @ 1.50GHz

which to me suggests "discrete GPU"

Henning

> 
> Matt
> 
> > 
> >  drivers/gpu/drm/i915/display/intel_bw.c | 80
> > +++--------------------- drivers/gpu/drm/i915/i915_drv.c         |
> > 4 ++ drivers/gpu/drm/i915/i915_drv.h         |  1 +
> >  drivers/gpu/drm/i915/i915_reg.h         |  3 +
> >  drivers/gpu/drm/i915/intel_dram.c       | 82
> > ++++++++++++++++++++++++- drivers/gpu/drm/i915/intel_sideband.c   |
> > 15 +++++ drivers/gpu/drm/i915/intel_sideband.h   |  2 +
> >  7 files changed, 114 insertions(+), 73 deletions(-)
> > 
> > -- 
> > 2.39.3
> > 
> >   
>