linux-kernel - Re: [PATCH] drm/amdgpu/gfx10: remove redundant repeated null checks

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <fda8103d-cac8-4c00-a78e-6eb27141d9ea@suswa.mountain>
Date: Mon, 4 Aug 2025 20:15:29 +0300
From: Dan Carpenter <dan.carpenter@...aro.org>
To: Alex Deucher <alexdeucher@...il.com>
Cc: Ethan Carter Edwards <ethan@...ancedwards.com>,
	Alex Deucher <alexander.deucher@....com>,
	Christian König <christian.koenig@....com>,
	David Airlie <airlied@...il.com>, Simona Vetter <simona@...ll.ch>,
	amd-gfx@...ts.freedesktop.org, dri-devel@...ts.freedesktop.org,
	linux-kernel@...r.kernel.org, kernel-janitors@...r.kernel.org
Subject: Re: [PATCH] drm/amdgpu/gfx10: remove redundant repeated null checks

On Mon, Aug 04, 2025 at 11:08:57AM -0400, Alex Deucher wrote:
> On Mon, Aug 4, 2025 at 10:49 AM Dan Carpenter <dan.carpenter@...aro.org> wrote:
> >
> > On Mon, Aug 04, 2025 at 10:32:43AM -0400, Alex Deucher wrote:
> > > On Sat, Aug 2, 2025 at 4:22 AM Ethan Carter Edwards
> > > <ethan@...ancedwards.com> wrote:
> > > >
> > > > The repeated checks on grbm_soft_reset are unnecessary. Remove them.
> > > >
> > >
> > > These are not NULL checks and they are necessary.  The code is
> > > checking if any bits are set in that register.  If not, then we can
> > > skip that code as there is nothing to do.
> > >
> >
> > It's not a null check, but it is a nested check and it's a local
> > variable so the patch is correct enough.  At this point we know that
> > grbm_soft_reset can't be zero.
> 
> It can be 0 as far as I can see.  If none of the GRBM_STATUS bits are
> set, then we never set any of the bits in grbm_soft_reset.
> 

You're missing the first check...

drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c
  7657          if (grbm_soft_reset) {
                    ^^^^^^^^^^^^^^^
Checked.

  7658                  /* stop the rlc */
  7659                  gfx_v10_0_rlc_stop(adev);
  7660  
  7661                  /* Disable GFX parsing/prefetching */
  7662                  gfx_v10_0_cp_gfx_enable(adev, false);
  7663  
  7664                  /* Disable MEC parsing/prefetching */
  7665                  gfx_v10_0_cp_compute_enable(adev, false);
  7666  
  7667                  if (grbm_soft_reset) {
                            ^^^^^^^^^^^^^^^
Unnecessary.

  7668                          tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
  7669                          tmp |= grbm_soft_reset;
  7670                          dev_info(adev->dev, "GRBM_SOFT_RESET=0x%08X\n", tmp);
  7671                          WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
  7672                          tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
  7673  
  7674                          udelay(50);
  7675  
  7676                          tmp &= ~grbm_soft_reset;
  7677                          WREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET, tmp);
  7678                          tmp = RREG32_SOC15(GC, 0, mmGRBM_SOFT_RESET);
  7679                  }
  7680  
  7681                  /* Wait a little for things to settle down */
  7682                  udelay(50);
  7683          }
  7684          return 0;

regards,
dan carpenter