AMD-GFX Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Alex Deucher <alexander.deucher@amd.com>
To: <amd-gfx@lists.freedesktop.org>
Cc: Alex Deucher <alexander.deucher@amd.com>
Subject: [PATCH 1/7] drm/amdgpu/gmc9: don't touch gfxhub registers during S0ix
Date: Thu, 15 Dec 2022 17:10:53 -0500	[thread overview]
Message-ID: <20221215221059.509565-2-alexander.deucher@amd.com> (raw)
In-Reply-To: <20221215221059.509565-1-alexander.deucher@amd.com>

gfxhub registers are part of gfx IP and should not need to be
changed.  Doing so without disabling gfxoff can hang the gfx IP.

v2: add comments explaining why we can skip the interrupt
    control for S0i3

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c | 36 ++++++++++++++++++++++-----
 1 file changed, 30 insertions(+), 6 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index 08d6cf79fb15..8f7fa468decb 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -484,6 +484,14 @@ static int gmc_v9_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
 			for (i = 0; i < 16; i++) {
 				reg = hub->vm_context0_cntl + i;
 
+				/* This works because this interrupt is only
+				 * enabled at init/resume and disabled in
+				 * fini/suspend, so the overall state doesn't
+				 * change over the course of suspend/resume.
+				 */
+				if (adev->in_s0ix && (j == AMDGPU_GFXHUB_0))
+					continue;
+
 				if (j == AMDGPU_GFXHUB_0)
 					tmp = RREG32_SOC15_IP(GC, reg);
 				else
@@ -504,6 +512,14 @@ static int gmc_v9_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
 			for (i = 0; i < 16; i++) {
 				reg = hub->vm_context0_cntl + i;
 
+				/* This works because this interrupt is only
+				 * enabled at init/resume and disabled in
+				 * fini/suspend, so the overall state doesn't
+				 * change over the course of suspend/resume.
+				 */
+				if (adev->in_s0ix && (j == AMDGPU_GFXHUB_0))
+					continue;
+
 				if (j == AMDGPU_GFXHUB_0)
 					tmp = RREG32_SOC15_IP(GC, reg);
 				else
@@ -1862,9 +1878,12 @@ static int gmc_v9_0_gart_enable(struct amdgpu_device *adev)
 	}
 
 	amdgpu_gtt_mgr_recover(&adev->mman.gtt_mgr);
-	r = adev->gfxhub.funcs->gart_enable(adev);
-	if (r)
-		return r;
+
+	if (!adev->in_s0ix) {
+		r = adev->gfxhub.funcs->gart_enable(adev);
+		if (r)
+			return r;
+	}
 
 	r = adev->mmhub.funcs->gart_enable(adev);
 	if (r)
@@ -1911,11 +1930,15 @@ static int gmc_v9_0_hw_init(void *handle)
 		value = true;
 
 	if (!amdgpu_sriov_vf(adev)) {
-		adev->gfxhub.funcs->set_fault_enable_default(adev, value);
+		if (!adev->in_s0ix)
+			adev->gfxhub.funcs->set_fault_enable_default(adev, value);
 		adev->mmhub.funcs->set_fault_enable_default(adev, value);
 	}
-	for (i = 0; i < adev->num_vmhubs; ++i)
+	for (i = 0; i < adev->num_vmhubs; ++i) {
+		if (adev->in_s0ix && (i == AMDGPU_GFXHUB_0))
+			continue;
 		gmc_v9_0_flush_gpu_tlb(adev, 0, i, 0);
+	}
 
 	if (adev->umc.funcs && adev->umc.funcs->init_registers)
 		adev->umc.funcs->init_registers(adev);
@@ -1939,7 +1962,8 @@ static int gmc_v9_0_hw_init(void *handle)
  */
 static void gmc_v9_0_gart_disable(struct amdgpu_device *adev)
 {
-	adev->gfxhub.funcs->gart_disable(adev);
+	if (!adev->in_s0ix)
+		adev->gfxhub.funcs->gart_disable(adev);
 	adev->mmhub.funcs->gart_disable(adev);
 }
 
-- 
2.38.1


  reply	other threads:[~2022-12-15 22:11 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-12-15 22:10 [PATCH 0/7 v2] Improve S0ix stability Alex Deucher
2022-12-15 22:10 ` Alex Deucher [this message]
2022-12-15 22:10 ` [PATCH 2/7] drm/amdgpu/gmc10: don't touch gfxhub registers during S0ix Alex Deucher
2022-12-15 22:10 ` [PATCH 3/7] drm/amdgpu/gmc11: " Alex Deucher
2022-12-15 22:10 ` [PATCH 4/7] drm/amdgpu: don't mess with SDMA clock or powergating in S0ix Alex Deucher
2022-12-15 22:10 ` [PATCH 5/7] drm/amdgpu: for S0ix, skip SMDA 5.x+ suspend/resume Alex Deucher
2022-12-16 14:35   ` [5/7] " Limonciello, Mario
2022-12-16 16:51     ` Alex Deucher
2022-12-16 15:15   ` [PATCH 5/7] " Russell, Kent
2022-12-15 22:10 ` [PATCH 6/7] Revert "drm/amdgpu: disallow gfxoff until GC IP blocks complete s2idle resume" Alex Deucher
2022-12-15 22:10 ` [PATCH 7/7] Revert "drm/amdgpu: force exit gfxoff on sdma resume for rmb s0ix" Alex Deucher
2022-12-16 14:23 ` [PATCH 0/7 v2] Improve S0ix stability Limonciello, Mario
  -- strict thread matches above, loose matches on Subject: below --
2022-12-14 22:16 [PATCH 0/7] Improvde " Alex Deucher
2022-12-14 22:16 ` [PATCH 1/7] drm/amdgpu/gmc9: don't touch gfxhub registers during S0ix Alex Deucher
2022-12-15  7:45   ` Christian König
2022-12-15 22:08     ` Alex Deucher

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20221215221059.509565-2-alexander.deucher@amd.com \
    --to=alexander.deucher@amd.com \
    --cc=amd-gfx@lists.freedesktop.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox