Re: Hard kernel lock with gentoo-sources-6.18.18

On Thu, Mar 26, 2026 at 4:39 AM Michael <[email protected]> wrote:
>
> On Wednesday, 25 March 2026 23:31:29 Greenwich Mean Time D. Ben Knoble
> wrote:
> > On Wed, Mar 25, 2026 at 2:25 PM Michael <[email protected]>
> wrote:
> > > On Tuesday, 24 March 2026 20:23:40 Greenwich Mean Time Tsukasa
> > >
> > > Mcp_Reznor wrote:
> > > > From: Michael
> > > > Sent: Monday, March 23, 2026 7:24 PM
> > > >
> > > > >I thought AMDGPU is stable code for many years, or at least it
> > > > >has
> > > >
> > > > >been with my card:
> > > > Lol, pretty sure the amdgpu drivers are like 1/4th the size of the
> > > > kernel, and they receive many many updates each kernel version,
> > > > albeit most of it will be for adding features or supporting the
> > > > latest and greatest.  With the shear scale of millions of lines of
> > > > code, it's actually fairly common for numerous new bugs each
> > > > release.
> > >
> > > I must have been lucky, but yet again I don't do any gaming.  I have
> > > not had such a glitch before over numerous desktops and laptops
> > > with Radeon cards, well since the good ol' fglrx days.  ;-)
> > >
> > > >  I've been rocking amd cards for decades exclusively, skipping a
> > > >
> > > > version here or there has always been worth it to me.  If you
> > > > check
> > > > out bugs.gentoo.org or use the kernels bug trackers, there will
> > > > probably already be a report on it, if not, feel free to add one,
> > > > in
> > > > the past I've been very impressed by gentoo's team, but this isn't
> > > > a
> > > > packaging or gentoo related bug most likely.  o7
> > >
> > > Hmm ... you are right.  The kernel driver appears to have have had a
> > > turn for the worse :
> > >
> > > https://community.frame.work/t/attn-critical-bugs-in-amdgpu-driver-i
> > > ncluded-with-kernel-6-18-x-6-19-x/79221
> > >
> > > I tried 6.19.9 and it went sideways in the same manner.  :-(
> >
> > Hm. I'm run 6.18.12 without issue right now, and my next reboot will
> > be for 6.18.18.
> >
> > I have an AMD RYZEN AI MAX+ 395 w/ Radeon 8060S from Framework; I
> > suppose I might have misunderstood what we mean by "amdgpu" but things
> > seem smooth here.
>
> I have two other older AMD A10 APUs (desktop & laptop) which work
> smoother than ever with the 6.18.18 and the amdgpu driver.  It's the
> Radeon RX Vega discrete GPU which has this problem.
>
> The "amdgpu" refers to this kernel driver:
>
> CONFIG_DRM_AMDGPU=y

Thanks; I have that as "m" and "lsmod | grep amdgpu" says

    amdgpu              18759680  29
    amdxcp                 12288  1 amdgpu
    drm_panel_backlight_quirks    12288  1 amdgpu
    drm_ttm_helper         12288  2 amdgpu
    ttm                   122880  2 amdgpu,drm_ttm_helper
    i2c_algo_bit           16384  1 amdgpu
    drm_suballoc_helper    16384  1 amdgpu
    drm_client_lib         12288  1 amdgpu
    drm_exec               12288  1 amdgpu
    drm_buddy              32768  1 amdgpu
    drm_display_helper    290816  1 amdgpu
    gpu_sched              61440  2 amdxdna,amdgpu
    cec                    94208  2 drm_display_helper,amdgpu
    drm_kms_helper        233472  5
drm_shmem_helper,drm_display_helper,amdgpu,drm_ttm_helper,drm_client_lib
    video                  73728  1 amdgpu

> Kernel 6.18.12 and prior have been faultless on all my PCs.  Kernels
> 6.18.18 and 6.19.9 both crash on the Vega.

I'm planning to reboot today, so I guess we'll see.

-- 
D. Ben Knoble