Skip to contentSkip to conversation

Patch "drm/nouveau/mmu/nv4a: use nv04 mmu rather than the nv44 one" has been added to the 4.9-stable tree

PATCH [email protected] 2 participants 9 messages May 22, 2017 – Jun 12, 2017

Subsystems (4)
lore
Share & tools
lore

[email protected]

author maintainer Applied
#0 lore raw

This is a note to let you know that I've just added the patch titled

drm/nouveau/mmu/nv4a: use nv04 mmu rather than the nv44 one

to the 4.9-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary

The filename of the patch is:

     drm-nouveau-mmu-nv4a-use-nv04-mmu-rather-than-the-nv44-one.patch

and it can be found in the queue-4.9 subdirectory.

If you, or anyone else, feels it should not be added to the stable tree, please let <[email protected]> know about it.

Quoted 1 line

From ac799acaa4d8db4f7dcd968b15c9596c80a4677f Mon Sep 17 00:00:00 2001

From: Ilia Mirkin <[email protected]>
Date: Sat, 18 Mar 2017 16:23:10 -0400
Subject: drm/nouveau/mmu/nv4a: use nv04 mmu rather than the nv44 one

From: Ilia Mirkin <[email protected]>

commit ac799acaa4d8db4f7dcd968b15c9596c80a4677f upstream.

The NV4A (aka NV44A) is an oddity in the family. It only comes in AGP and PCI varieties, rather than a core PCIE chip with a bridge for AGP/PCI as necessary. As a result, it appears that the MMU is also non-functional. For AGP cards, the vast majority of the NV4A lineup, this worked out since we force AGP cards to use the nv04 mmu. However for PCI variants, this did not work.

Switching to the NV04 MMU makes it work like a charm. Thanks to mwk for the suggestion. This should be a no-op for NV4A AGP boards, as they were using it already.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=70388

---
 drivers/gpu/drm/nouveau/nvkm/engine/device/base.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/nouveau/nvkm/engine/device/base.c +1 -1
@@ -584,7 +584,7 @@ nv44_chipset = {
 	.i2c = nv04_i2c_new,
 	.imem = nv40_instmem_new,
 	.mc = nv44_mc_new,
-	.mmu = nv44_mmu_new,
+	.mmu = nv04_mmu_new,
 	.pci = nv40_pci_new,
 	.therm = nv40_therm_new,
 	.timer = nv41_timer_new,

Patches currently in stable-queue which might be from [email protected] are

queue-4.9/drm-nouveau-mmu-nv4a-use-nv04-mmu-rather-than-the-nv44-one.patch

#1 lore raw

Hi Greg,

This patch appears to have misapplied. I'm not sure why cherry-pick got it wrong, but this is meant to go in the nv4a_chipset definition, not nv44_chipset.

This comment also applies to the 4.4 and 4.11 cherry-picks you just did.

Thanks,

-ilia

Quoted 61 lines · On Mon, May 22, 2017 at 1:04 PM, <[email protected]> wrote:

This is a note to let you know that I've just added the patch titled

drm/nouveau/mmu/nv4a: use nv04 mmu rather than the nv44 one

to the 4.9-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary

The filename of the patch is:

     drm-nouveau-mmu-nv4a-use-nv04-mmu-rather-than-the-nv44-one.patch

and it can be found in the queue-4.9 subdirectory.

If you, or anyone else, feels it should not be added to the stable tree, please let <[email protected]> know about it.

From ac799acaa4d8db4f7dcd968b15c9596c80a4677f Mon Sep 17 00:00:00 2001
From: Ilia Mirkin <[email protected]>
Date: Sat, 18 Mar 2017 16:23:10 -0400
Subject: drm/nouveau/mmu/nv4a: use nv04 mmu rather than the nv44 one

From: Ilia Mirkin <[email protected]>

commit ac799acaa4d8db4f7dcd968b15c9596c80a4677f upstream.

The NV4A (aka NV44A) is an oddity in the family. It only comes in AGP and PCI varieties, rather than a core PCIE chip with a bridge for AGP/PCI as necessary. As a result, it appears that the MMU is also non-functional. For AGP cards, the vast majority of the NV4A lineup, this worked out since we force AGP cards to use the nv04 mmu. However for PCI variants, this did not work.

Switching to the NV04 MMU makes it work like a charm. Thanks to mwk for the suggestion. This should be a no-op for NV4A AGP boards, as they were using it already.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=70388

---
 drivers/gpu/drm/nouveau/nvkm/engine/device/base.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/nouveau/nvkm/engine/device/base.c +1 -1
@@ -584,7 +584,7 @@ nv44_chipset = {
        .i2c = nv04_i2c_new,
        .imem = nv40_instmem_new,
        .mc = nv44_mc_new,
-       .mmu = nv44_mmu_new,
+       .mmu = nv04_mmu_new,
        .pci = nv40_pci_new,
        .therm = nv40_therm_new,
        .timer = nv41_timer_new,

Patches currently in stable-queue which might be from [email protected] are

queue-4.9/drm-nouveau-mmu-nv4a-use-nv04-mmu-rather-than-the-nv44-one.patch

#2 lore raw
Quoted 7 lines · Ilia Mirkin

Hi Greg,

This patch appears to have misapplied. I'm not sure why cherry-pick got it wrong, but this is meant to go in the nv4a_chipset definition, not nv44_chipset.

This comment also applies to the 4.4 and 4.11 cherry-picks you just did.

Aha, I see what happened. f94773b9f5ecd1df7c88c2e921924dd41d2020cc made it into v4.11. However by then it appears to have also made it into drm-next as ac799acaa4d8db4f7dcd968b15c9596c80a4677f and thus the drm merge into v4.12. So now you have two versions of the same commit and are trying to cherry-pick both.

I'm guessing it's also the reason why your cherrypick of my other commit failed (ad01a91a820cd9f0d880c407bf556b67298dcc93).

And I might note that "drm-nouveau-mmu-nv4a-use-nv04-mmu-rather-than-the-nv44-one.patch" was already there before, so didn't make it into your "4.9-stable patches" commit on the stable-queue git tree.

-ilia

Quoted 67 lines

Thanks,

-ilia

Quoted 61 lines · On Mon, May 22, 2017 at 1:04 PM, <[email protected]> wrote:

This is a note to let you know that I've just added the patch titled

drm/nouveau/mmu/nv4a: use nv04 mmu rather than the nv44 one

to the 4.9-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary

The filename of the patch is:

     drm-nouveau-mmu-nv4a-use-nv04-mmu-rather-than-the-nv44-one.patch

and it can be found in the queue-4.9 subdirectory.

If you, or anyone else, feels it should not be added to the stable tree, please let <[email protected]> know about it.

From ac799acaa4d8db4f7dcd968b15c9596c80a4677f Mon Sep 17 00:00:00 2001
From: Ilia Mirkin <[email protected]>
Date: Sat, 18 Mar 2017 16:23:10 -0400
Subject: drm/nouveau/mmu/nv4a: use nv04 mmu rather than the nv44 one

From: Ilia Mirkin <[email protected]>

commit ac799acaa4d8db4f7dcd968b15c9596c80a4677f upstream.

The NV4A (aka NV44A) is an oddity in the family. It only comes in AGP and PCI varieties, rather than a core PCIE chip with a bridge for AGP/PCI as necessary. As a result, it appears that the MMU is also non-functional. For AGP cards, the vast majority of the NV4A lineup, this worked out since we force AGP cards to use the nv04 mmu. However for PCI variants, this did not work.

Switching to the NV04 MMU makes it work like a charm. Thanks to mwk for the suggestion. This should be a no-op for NV4A AGP boards, as they were using it already.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=70388

---
 drivers/gpu/drm/nouveau/nvkm/engine/device/base.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/nouveau/nvkm/engine/device/base.c +1 -1
@@ -584,7 +584,7 @@ nv44_chipset = {
        .i2c = nv04_i2c_new,
        .imem = nv40_instmem_new,
        .mc = nv44_mc_new,
-       .mmu = nv44_mmu_new,
+       .mmu = nv04_mmu_new,
        .pci = nv40_pci_new,
        .therm = nv40_therm_new,
        .timer = nv41_timer_new,

Patches currently in stable-queue which might be from [email protected] are

queue-4.9/drm-nouveau-mmu-nv4a-use-nv04-mmu-rather-than-the-nv44-one.patch

author maintainer
#3 lore raw
Quoted 7 lines · Ilia Mirkin
Quoted 6 lines · Ilia Mirkin

Hi Greg,

This patch appears to have misapplied. I'm not sure why cherry-pick got it wrong, but this is meant to go in the nv4a_chipset definition, not nv44_chipset.

Quoted 7 lines
Quoted 1 line

This comment also applies to the 4.4 and 4.11 cherry-picks you just did.

Aha, I see what happened. f94773b9f5ecd1df7c88c2e921924dd41d2020cc made it into v4.11. However by then it appears to have also made it into drm-next as ac799acaa4d8db4f7dcd968b15c9596c80a4677f and thus the drm merge into v4.12. So now you have two versions of the same commit and are trying to cherry-pick both.

Why in the world would you include the same patch in two different trees with two different git ids? I'm _really_ starting to hate the drm trees...

Quoted 2 lines

I'm guessing it's also the reason why your cherrypick of my other commit failed (ad01a91a820cd9f0d880c407bf556b67298dcc93).

That would make sense, but again, why is this happening?

Quoted 4 lines

And I might note that "drm-nouveau-mmu-nv4a-use-nv04-mmu-rather-than-the-nv44-one.patch" was already there before, so didn't make it into your "4.9-stable patches" commit on the stable-queue git tree.

Ok, so what should i do here? Drop all of the nouveau patches? Some of them? Which ones were wrong?

totally confused and grumpy,

greg k-h

#4 lore raw
Quoted 16 lines · Greg Kroah-Hartman
Quoted 7 lines · Ilia Mirkin
Quoted 6 lines · Ilia Mirkin

Hi Greg,

This patch appears to have misapplied. I'm not sure why cherry-pick got it wrong, but this is meant to go in the nv4a_chipset definition, not nv44_chipset.

Quoted 7 lines
Quoted 1 line

This comment also applies to the 4.4 and 4.11 cherry-picks you just did.

Aha, I see what happened. f94773b9f5ecd1df7c88c2e921924dd41d2020cc made it into v4.11. However by then it appears to have also made it into drm-next as ac799acaa4d8db4f7dcd968b15c9596c80a4677f and thus the drm merge into v4.12. So now you have two versions of the same commit and are trying to cherry-pick both.

Quoted 3 lines

Why in the world would you include the same patch in two different trees with two different git ids? I'm _really_ starting to hate the drm trees...

This question may be above my pay-grade (I'm just a lowly occasional contributor), but I believe the situation is that first the change goes into drm-next, and then is cherry-picked into drm-fixes. I haven't checked that this is what happened though. Nor am I sufficiently plugged into linux kernel development processes to know what the proper way to handle that situation would be, if any. Something to address with Ben Skeggs and Dave Airlie, I suppose.

Quoted 6 lines
Quoted 3 lines

I'm guessing it's also the reason why your cherrypick of my other commit failed (ad01a91a820cd9f0d880c407bf556b67298dcc93).

That would make sense, but again, why is this happening?

Quoted 7 lines
Quoted 4 lines

And I might note that "drm-nouveau-mmu-nv4a-use-nv04-mmu-rather-than-the-nv44-one.patch" was already there before, so didn't make it into your "4.9-stable patches" commit on the stable-queue git tree.

Ok, so what should i do here? Drop all of the nouveau patches? Some of them? Which ones were wrong?

Looking at e.g.

https://git.kernel.org/pub/scm/linux/kernel/git/stable/stable-queue.git/tree/queue-4.9

I can say unequivocally that "drm-nouveau-mmu-nv4a-use-nv04-mmu-rather-than-the-nv44-one.patch" is wrong in all three of your queues (4.4, 4.9, 4.11). Please drop it.

AFAIK the others should be good, but they're not my patches.

Cheers,

-ilia

author maintainer
#5 lore raw
Quoted 22 lines · Ilia Mirkin
Quoted 16 lines · Greg Kroah-Hartman
Quoted 7 lines · Ilia Mirkin
Quoted 6 lines · Ilia Mirkin

Hi Greg,

This patch appears to have misapplied. I'm not sure why cherry-pick got it wrong, but this is meant to go in the nv4a_chipset definition, not nv44_chipset.

Quoted 7 lines
Quoted 1 line

This comment also applies to the 4.4 and 4.11 cherry-picks you just did.

Aha, I see what happened. f94773b9f5ecd1df7c88c2e921924dd41d2020cc made it into v4.11. However by then it appears to have also made it into drm-next as ac799acaa4d8db4f7dcd968b15c9596c80a4677f and thus the drm merge into v4.12. So now you have two versions of the same commit and are trying to cherry-pick both.

Quoted 3 lines

Why in the world would you include the same patch in two different trees with two different git ids? I'm _really_ starting to hate the drm trees...

Quoted 7 lines

This question may be above my pay-grade (I'm just a lowly occasional contributor), but I believe the situation is that first the change goes into drm-next, and then is cherry-picked into drm-fixes. I haven't checked that this is what happened though. Nor am I sufficiently plugged into linux kernel development processes to know what the proper way to handle that situation would be, if any. Something to address with Ben Skeggs and Dave Airlie, I suppose.

It seems that the i915 model of putting duplicate patches in different branches is seeping over into other drm driver trees :(

The people involved with that know my position there, it's a pain in the ass for people like me, for reasons exactly like you are seeing here. Thanks for helping out with this.

Quoted 13 lines
Quoted 5 lines
Quoted 2 lines

I'm guessing it's also the reason why your cherrypick of my other commit failed (ad01a91a820cd9f0d880c407bf556b67298dcc93).

That would make sense, but again, why is this happening?

Quoted 7 lines
Quoted 4 lines

And I might note that "drm-nouveau-mmu-nv4a-use-nv04-mmu-rather-than-the-nv44-one.patch" was already there before, so didn't make it into your "4.9-stable patches" commit on the stable-queue git tree.

Ok, so what should i do here? Drop all of the nouveau patches? Some of them? Which ones were wrong?

Quoted 7 lines

Looking at e.g.

https://git.kernel.org/pub/scm/linux/kernel/git/stable/stable-queue.git/tree/queue-4.9

I can say unequivocally that "drm-nouveau-mmu-nv4a-use-nv04-mmu-rather-than-the-nv44-one.patch" is wrong in all three of your queues (4.4, 4.9, 4.11). Please drop it.

Ok, now all dropped.

thanks again for catching this and letting me know what to do,

greg k-h

author maintainer
#6 lore raw
In reply to Greg Kroah-Hartman
Quoted 30 lines · Greg Kroah-Hartman
Quoted 7 lines · Ilia Mirkin
Quoted 6 lines · Ilia Mirkin

Hi Greg,

This patch appears to have misapplied. I'm not sure why cherry-pick got it wrong, but this is meant to go in the nv4a_chipset definition, not nv44_chipset.

Quoted 7 lines
Quoted 1 line

This comment also applies to the 4.4 and 4.11 cherry-picks you just did.

Aha, I see what happened. f94773b9f5ecd1df7c88c2e921924dd41d2020cc made it into v4.11. However by then it appears to have also made it into drm-next as ac799acaa4d8db4f7dcd968b15c9596c80a4677f and thus the drm merge into v4.12. So now you have two versions of the same commit and are trying to cherry-pick both.

Why in the world would you include the same patch in two different trees with two different git ids? I'm _really_ starting to hate the drm trees...

Quoted 2 lines

I'm guessing it's also the reason why your cherrypick of my other commit failed (ad01a91a820cd9f0d880c407bf556b67298dcc93).

That would make sense, but again, why is this happening?

Quoted 4 lines

And I might note that "drm-nouveau-mmu-nv4a-use-nv04-mmu-rather-than-the-nv44-one.patch" was already there before, so didn't make it into your "4.9-stable patches" commit on the stable-queue git tree.

Quoted 4 lines

Ok, so what should i do here? Drop all of the nouveau patches? Some of them? Which ones were wrong?

totally confused and grumpy,

Still confused, I guess all is ok in the stable trees? Not according to some bug reports we are getting, 4.4 seems broken, oh well, hopefully everyone is using newer kernel releases...

if you still want me to do something here, please let me know.

thanks,

greg k-h

#7 lore raw
Quoted 31 lines · Greg Kroah-Hartman
Quoted 30 lines · Greg Kroah-Hartman
Quoted 7 lines · Ilia Mirkin
Quoted 6 lines · Ilia Mirkin

Hi Greg,

This patch appears to have misapplied. I'm not sure why cherry-pick got it wrong, but this is meant to go in the nv4a_chipset definition, not nv44_chipset.

Quoted 7 lines
Quoted 1 line

This comment also applies to the 4.4 and 4.11 cherry-picks you just did.

Aha, I see what happened. f94773b9f5ecd1df7c88c2e921924dd41d2020cc made it into v4.11. However by then it appears to have also made it into drm-next as ac799acaa4d8db4f7dcd968b15c9596c80a4677f and thus the drm merge into v4.12. So now you have two versions of the same commit and are trying to cherry-pick both.

Why in the world would you include the same patch in two different trees with two different git ids? I'm _really_ starting to hate the drm trees...

Quoted 2 lines

I'm guessing it's also the reason why your cherrypick of my other commit failed (ad01a91a820cd9f0d880c407bf556b67298dcc93).

That would make sense, but again, why is this happening?

Quoted 4 lines

And I might note that "drm-nouveau-mmu-nv4a-use-nv04-mmu-rather-than-the-nv44-one.patch" was already there before, so didn't make it into your "4.9-stable patches" commit on the stable-queue git tree.

Quoted 7 lines
Quoted 4 lines

Ok, so what should i do here? Drop all of the nouveau patches? Some of them? Which ones were wrong?

totally confused and grumpy,

Still confused, I guess all is ok in the stable trees? Not according to some bug reports we are getting, 4.4 seems broken,

Pointer? I'd be highly surprised if anyone were to run into the issue that my patch was addressing ("using old hardware"), or that my fix had caused problems for anyone. However I can confirm that the MMU pointer is set properly in the 4.4 tree for both the NV44 and NV44A chips.

Note that there's a totally unrelated issue in the patch series ending
in https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git/commit/drivers/gpu/drm/nouveau?h=linux-4.4.y&id=e4add1cf6b4154804350c3385c6d447cff3570de
but Ben is aware of that issue and has a patch. I suspect it should be
making its way to stable-land ... eventually? On the bright side, that
code was broken before too and led to kernel thread deadlocks on some
hardware.

Cheers,

-ilia

author maintainer
#8 lore raw
Quoted 38 lines · Ilia Mirkin
Quoted 36 lines · Greg Kroah-Hartman
Quoted 34 lines · Greg Kroah-Hartman
Quoted 7 lines · Ilia Mirkin
Quoted 6 lines · Ilia Mirkin

Hi Greg,

This patch appears to have misapplied. I'm not sure why cherry-pick got it wrong, but this is meant to go in the nv4a_chipset definition, not nv44_chipset.

Quoted 7 lines
Quoted 1 line

This comment also applies to the 4.4 and 4.11 cherry-picks you just did.

Aha, I see what happened. f94773b9f5ecd1df7c88c2e921924dd41d2020cc made it into v4.11. However by then it appears to have also made it into drm-next as ac799acaa4d8db4f7dcd968b15c9596c80a4677f and thus the drm merge into v4.12. So now you have two versions of the same commit and are trying to cherry-pick both.

Why in the world would you include the same patch in two different trees with two different git ids? I'm _really_ starting to hate the drm trees...

Quoted 2 lines

I'm guessing it's also the reason why your cherrypick of my other commit failed (ad01a91a820cd9f0d880c407bf556b67298dcc93).

That would make sense, but again, why is this happening?

Quoted 4 lines

And I might note that "drm-nouveau-mmu-nv4a-use-nv04-mmu-rather-than-the-nv44-one.patch" was already there before, so didn't make it into your "4.9-stable patches" commit on the stable-queue git tree.

Ok, so what should i do here? Drop all of the nouveau patches? Some of them? Which ones were wrong?

totally confused and grumpy,

Quoted 8 lines
Quoted 2 lines

Still confused, I guess all is ok in the stable trees? Not according to some bug reports we are getting, 4.4 seems broken,

Pointer? I'd be highly surprised if anyone were to run into the issue that my patch was addressing ("using old hardware"), or that my fix had caused problems for anyone. However I can confirm that the MMU pointer is set properly in the 4.4 tree for both the NV44 and NV44A chips.

See:
Subject: [4.4.70 REGRESSION] Nouveau hangs up at boot

posted to lkml for details.

thanks,

greg k-h