| Message ID | 20190521161102.29620-1-peron.clem@gmail.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | Allwinner H6 Mali GPU support | expand |
On Tue, May 21, 2019 at 11:11 AM Clément Péron <peron.clem@gmail.com> wrote: > > Hi, > > The Allwinner H6 has a Mali-T720 MP2 which should be supported by > the new panfrost driver. This series fix two issues and introduce the > dt-bindings but a simple benchmark show that it's still NOT WORKING. > > I'm pushing it in case someone want to continue the work. > > This has been tested with Mesa3D 19.1.0-RC2 and a GPU bitness patch[1]. > > One patch is from Icenowy Zheng where I changed the order as required > by Rob Herring[2]. > > Thanks, > Clement > > [1] https://gitlab.freedesktop.org/kszaq/mesa/tree/panfrost_64_32 > [2] https://patchwork.kernel.org/patch/10699829/ > > > [ 345.204813] panfrost 1800000.gpu: mmu irq status=1 > [ 345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > 0x0000000002400400 > [ 345.209617] Reason: TODO > [ 345.209617] raw fault status: 0x800002C1 > [ 345.209617] decoded fault status: SLAVE FAULT > [ 345.209617] exception type 0xC1: TRANSLATION_FAULT_LEVEL1 > [ 345.209617] access type 0x2: READ > [ 345.209617] source id 0x8000 > [ 345.729957] panfrost 1800000.gpu: gpu sched timeout, js=0, > status=0x8, head=0x2400400, tail=0x2400400, sched_job=000000009e204de9 > [ 346.055876] panfrost 1800000.gpu: mmu irq status=1 > [ 346.060680] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > 0x0000000002C00A00 > [ 346.060680] Reason: TODO > [ 346.060680] raw fault status: 0x810002C1 > [ 346.060680] decoded fault status: SLAVE FAULT > [ 346.060680] exception type 0xC1: TRANSLATION_FAULT_LEVEL1 > [ 346.060680] access type 0x2: READ > [ 346.060680] source id 0x8100 > [ 346.561955] panfrost 1800000.gpu: gpu sched timeout, js=1, > status=0x8, head=0x2c00a00, tail=0x2c00a00, sched_job=00000000b55a9a85 > [ 346.573913] panfrost 1800000.gpu: mmu irq status=1 > [ 346.578707] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > 0x0000000002C00B80 > > Change in v5: > - Remove fix indent > > Changes in v4: > - Add bus_clock probe > - Fix sanity check in io-pgtable > - Add vramp-delay > - Merge all boards into one patch > - Remove upstreamed Neil A. patch > > Change in v3 (Thanks to Maxime Ripard): > - Reauthor Icenowy for her path > > Changes in v2 (Thanks to Maxime Ripard): > - Drop GPU OPP Table > - Add clocks and clock-names in required > > Clément Péron (5): > drm: panfrost: add optional bus_clock > iommu: io-pgtable: fix sanity check for non 48-bit mali iommu > dt-bindings: gpu: mali-midgard: Add H6 mali gpu compatible > arm64: dts: allwinner: Add ARM Mali GPU node for H6 > arm64: dts: allwinner: Add mali GPU supply for H6 boards > > Icenowy Zheng (1): > dt-bindings: gpu: add bus clock for Mali Midgard GPUs I've applied patches 1 and 3 to drm-misc. I was going to do patch 4 too, but it doesn't apply. Patch 2 can go in via the iommu tree and the rest via the allwinner tree. Rob
Hi Rob, On Wed, 22 May 2019 at 21:27, Rob Herring <robh+dt@kernel.org> wrote: > > On Tue, May 21, 2019 at 11:11 AM Clément Péron <peron.clem@gmail.com> wrote: > > > > Hi, > > > > The Allwinner H6 has a Mali-T720 MP2 which should be supported by > > the new panfrost driver. This series fix two issues and introduce the > > dt-bindings but a simple benchmark show that it's still NOT WORKING. > > > > I'm pushing it in case someone want to continue the work. > > > > This has been tested with Mesa3D 19.1.0-RC2 and a GPU bitness patch[1]. > > > > One patch is from Icenowy Zheng where I changed the order as required > > by Rob Herring[2]. > > > > Thanks, > > Clement > > > > [1] https://gitlab.freedesktop.org/kszaq/mesa/tree/panfrost_64_32 > > [2] https://patchwork.kernel.org/patch/10699829/ > > > > > > [ 345.204813] panfrost 1800000.gpu: mmu irq status=1 > > [ 345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > > 0x0000000002400400 > > [ 345.209617] Reason: TODO > > [ 345.209617] raw fault status: 0x800002C1 > > [ 345.209617] decoded fault status: SLAVE FAULT > > [ 345.209617] exception type 0xC1: TRANSLATION_FAULT_LEVEL1 > > [ 345.209617] access type 0x2: READ > > [ 345.209617] source id 0x8000 > > [ 345.729957] panfrost 1800000.gpu: gpu sched timeout, js=0, > > status=0x8, head=0x2400400, tail=0x2400400, sched_job=000000009e204de9 > > [ 346.055876] panfrost 1800000.gpu: mmu irq status=1 > > [ 346.060680] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > > 0x0000000002C00A00 > > [ 346.060680] Reason: TODO > > [ 346.060680] raw fault status: 0x810002C1 > > [ 346.060680] decoded fault status: SLAVE FAULT > > [ 346.060680] exception type 0xC1: TRANSLATION_FAULT_LEVEL1 > > [ 346.060680] access type 0x2: READ > > [ 346.060680] source id 0x8100 > > [ 346.561955] panfrost 1800000.gpu: gpu sched timeout, js=1, > > status=0x8, head=0x2c00a00, tail=0x2c00a00, sched_job=00000000b55a9a85 > > [ 346.573913] panfrost 1800000.gpu: mmu irq status=1 > > [ 346.578707] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > > 0x0000000002C00B80 > > > > Change in v5: > > - Remove fix indent > > > > Changes in v4: > > - Add bus_clock probe > > - Fix sanity check in io-pgtable > > - Add vramp-delay > > - Merge all boards into one patch > > - Remove upstreamed Neil A. patch > > > > Change in v3 (Thanks to Maxime Ripard): > > - Reauthor Icenowy for her path > > > > Changes in v2 (Thanks to Maxime Ripard): > > - Drop GPU OPP Table > > - Add clocks and clock-names in required > > > > Clément Péron (5): > > drm: panfrost: add optional bus_clock > > iommu: io-pgtable: fix sanity check for non 48-bit mali iommu > > dt-bindings: gpu: mali-midgard: Add H6 mali gpu compatible > > arm64: dts: allwinner: Add ARM Mali GPU node for H6 > > arm64: dts: allwinner: Add mali GPU supply for H6 boards > > > > Icenowy Zheng (1): > > dt-bindings: gpu: add bus clock for Mali Midgard GPUs > > I've applied patches 1 and 3 to drm-misc. I was going to do patch 4 > too, but it doesn't apply. Thanks, I have tried to applied on drm-misc/for-linux-next but it doesn't apply too. It looks like commit d5ff1adb3809e2f74a3b57cea2e57c8e37d693c4 is missing on drm-misc ? https://github.com/torvalds/linux/commit/d5ff1adb3809e2f74a3b57cea2e57c8e37d693c4#diff-c3172f5d421d492ff91d7bb44dd44917 Clément > > Patch 2 can go in via the iommu tree and the rest via the allwinner tree. > > Rob
On Wed, May 22, 2019 at 2:41 PM Clément Péron <peron.clem@gmail.com> wrote: > > Hi Rob, > > On Wed, 22 May 2019 at 21:27, Rob Herring <robh+dt@kernel.org> wrote: > > > > On Tue, May 21, 2019 at 11:11 AM Clément Péron <peron.clem@gmail.com> wrote: > > > > > > Hi, > > > > > > The Allwinner H6 has a Mali-T720 MP2 which should be supported by > > > the new panfrost driver. This series fix two issues and introduce the > > > dt-bindings but a simple benchmark show that it's still NOT WORKING. > > > > > > I'm pushing it in case someone want to continue the work. > > > > > > This has been tested with Mesa3D 19.1.0-RC2 and a GPU bitness patch[1]. > > > > > > One patch is from Icenowy Zheng where I changed the order as required > > > by Rob Herring[2]. > > > > > > Thanks, > > > Clement > > > > > > [1] https://gitlab.freedesktop.org/kszaq/mesa/tree/panfrost_64_32 > > > [2] https://patchwork.kernel.org/patch/10699829/ > > > > > > > > > [ 345.204813] panfrost 1800000.gpu: mmu irq status=1 > > > [ 345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > > > 0x0000000002400400 > > > [ 345.209617] Reason: TODO > > > [ 345.209617] raw fault status: 0x800002C1 > > > [ 345.209617] decoded fault status: SLAVE FAULT > > > [ 345.209617] exception type 0xC1: TRANSLATION_FAULT_LEVEL1 > > > [ 345.209617] access type 0x2: READ > > > [ 345.209617] source id 0x8000 > > > [ 345.729957] panfrost 1800000.gpu: gpu sched timeout, js=0, > > > status=0x8, head=0x2400400, tail=0x2400400, sched_job=000000009e204de9 > > > [ 346.055876] panfrost 1800000.gpu: mmu irq status=1 > > > [ 346.060680] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > > > 0x0000000002C00A00 > > > [ 346.060680] Reason: TODO > > > [ 346.060680] raw fault status: 0x810002C1 > > > [ 346.060680] decoded fault status: SLAVE FAULT > > > [ 346.060680] exception type 0xC1: TRANSLATION_FAULT_LEVEL1 > > > [ 346.060680] access type 0x2: READ > > > [ 346.060680] source id 0x8100 > > > [ 346.561955] panfrost 1800000.gpu: gpu sched timeout, js=1, > > > status=0x8, head=0x2c00a00, tail=0x2c00a00, sched_job=00000000b55a9a85 > > > [ 346.573913] panfrost 1800000.gpu: mmu irq status=1 > > > [ 346.578707] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > > > 0x0000000002C00B80 > > > > > > Change in v5: > > > - Remove fix indent > > > > > > Changes in v4: > > > - Add bus_clock probe > > > - Fix sanity check in io-pgtable > > > - Add vramp-delay > > > - Merge all boards into one patch > > > - Remove upstreamed Neil A. patch > > > > > > Change in v3 (Thanks to Maxime Ripard): > > > - Reauthor Icenowy for her path > > > > > > Changes in v2 (Thanks to Maxime Ripard): > > > - Drop GPU OPP Table > > > - Add clocks and clock-names in required > > > > > > Clément Péron (5): > > > drm: panfrost: add optional bus_clock > > > iommu: io-pgtable: fix sanity check for non 48-bit mali iommu > > > dt-bindings: gpu: mali-midgard: Add H6 mali gpu compatible > > > arm64: dts: allwinner: Add ARM Mali GPU node for H6 > > > arm64: dts: allwinner: Add mali GPU supply for H6 boards > > > > > > Icenowy Zheng (1): > > > dt-bindings: gpu: add bus clock for Mali Midgard GPUs > > > > I've applied patches 1 and 3 to drm-misc. I was going to do patch 4 > > too, but it doesn't apply. > > Thanks, > > I have tried to applied on drm-misc/for-linux-next but it doesn't apply too. > It looks like commit d5ff1adb3809e2f74a3b57cea2e57c8e37d693c4 is > missing on drm-misc ? > https://github.com/torvalds/linux/commit/d5ff1adb3809e2f74a3b57cea2e57c8e37d693c4#diff-c3172f5d421d492ff91d7bb44dd44917 5.2-rc1 is merged in now and I've applied patch 4. Rob
On 21/05/2019 17:10, Clément Péron wrote: > Hi, > > The Allwinner H6 has a Mali-T720 MP2 which should be supported by > the new panfrost driver. This series fix two issues and introduce the > dt-bindings but a simple benchmark show that it's still NOT WORKING. > > I'm pushing it in case someone want to continue the work. > > This has been tested with Mesa3D 19.1.0-RC2 and a GPU bitness patch[1]. > > One patch is from Icenowy Zheng where I changed the order as required > by Rob Herring[2]. > > Thanks, > Clement > > [1] https://gitlab.freedesktop.org/kszaq/mesa/tree/panfrost_64_32 > [2] https://patchwork.kernel.org/patch/10699829/ > > > [ 345.204813] panfrost 1800000.gpu: mmu irq status=1 > [ 345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > 0x0000000002400400 > [ 345.209617] Reason: TODO > [ 345.209617] raw fault status: 0x800002C1 > [ 345.209617] decoded fault status: SLAVE FAULT > [ 345.209617] exception type 0xC1: TRANSLATION_FAULT_LEVEL1 > [ 345.209617] access type 0x2: READ > [ 345.209617] source id 0x8000 > [ 345.729957] panfrost 1800000.gpu: gpu sched timeout, js=0, > status=0x8, head=0x2400400, tail=0x2400400, sched_job=000000009e204de9 > [ 346.055876] panfrost 1800000.gpu: mmu irq status=1 > [ 346.060680] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > 0x0000000002C00A00 > [ 346.060680] Reason: TODO > [ 346.060680] raw fault status: 0x810002C1 > [ 346.060680] decoded fault status: SLAVE FAULT > [ 346.060680] exception type 0xC1: TRANSLATION_FAULT_LEVEL1 > [ 346.060680] access type 0x2: READ > [ 346.060680] source id 0x8100 > [ 346.561955] panfrost 1800000.gpu: gpu sched timeout, js=1, > status=0x8, head=0x2c00a00, tail=0x2c00a00, sched_job=00000000b55a9a85 > [ 346.573913] panfrost 1800000.gpu: mmu irq status=1 > [ 346.578707] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > 0x0000000002C00B80 FWIW I seem to have reproduced the same behaviour on a different T720 setup, so this may well be more about the GPU than the platform. There doesn't look to be anything obviously wrong with the pagetables, but if I can find some more free time I may have a bit more of a poke around. Robin. > > Change in v5: > - Remove fix indent > > Changes in v4: > - Add bus_clock probe > - Fix sanity check in io-pgtable > - Add vramp-delay > - Merge all boards into one patch > - Remove upstreamed Neil A. patch > > Change in v3 (Thanks to Maxime Ripard): > - Reauthor Icenowy for her path > > Changes in v2 (Thanks to Maxime Ripard): > - Drop GPU OPP Table > - Add clocks and clock-names in required > > Clément Péron (5): > drm: panfrost: add optional bus_clock > iommu: io-pgtable: fix sanity check for non 48-bit mali iommu > dt-bindings: gpu: mali-midgard: Add H6 mali gpu compatible > arm64: dts: allwinner: Add ARM Mali GPU node for H6 > arm64: dts: allwinner: Add mali GPU supply for H6 boards > > Icenowy Zheng (1): > dt-bindings: gpu: add bus clock for Mali Midgard GPUs > > .../bindings/gpu/arm,mali-midgard.txt | 15 ++++++++++++- > .../dts/allwinner/sun50i-h6-beelink-gs1.dts | 6 +++++ > .../dts/allwinner/sun50i-h6-orangepi-3.dts | 6 +++++ > .../dts/allwinner/sun50i-h6-orangepi.dtsi | 6 +++++ > .../boot/dts/allwinner/sun50i-h6-pine-h64.dts | 6 +++++ > arch/arm64/boot/dts/allwinner/sun50i-h6.dtsi | 14 ++++++++++++ > drivers/gpu/drm/panfrost/panfrost_device.c | 22 +++++++++++++++++++ > drivers/gpu/drm/panfrost/panfrost_device.h | 1 + > drivers/iommu/io-pgtable-arm.c | 2 +- > 9 files changed, 76 insertions(+), 2 deletions(-) >
On Tue, 21 May 2019 at 18:11, Clément Péron <peron.clem@gmail.com> wrote: > [snip] > [ 345.204813] panfrost 1800000.gpu: mmu irq status=1 > [ 345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > 0x0000000002400400 From what I can see here, 0x0000000002400400 points to the first byte of the first submitted job descriptor. So mapping buffers for the GPU doesn't seem to be working at all on 64-bit T-760. Steven, Robin, do you have any idea of why this could be? Thanks, Tomeu
On 29/05/2019 16:09, Tomeu Vizoso wrote: > On Tue, 21 May 2019 at 18:11, Clément Péron <peron.clem@gmail.com> wrote: >> > [snip] >> [ 345.204813] panfrost 1800000.gpu: mmu irq status=1 >> [ 345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA >> 0x0000000002400400 > > From what I can see here, 0x0000000002400400 points to the first byte > of the first submitted job descriptor. > > So mapping buffers for the GPU doesn't seem to be working at all on > 64-bit T-760. > > Steven, Robin, do you have any idea of why this could be? I tried rolling back to the old panfrost/nondrm shim, and it works fine with kbase, and I also found that T-820 falls over in the exact same manner, so the fact that it seemed to be common to the smaller 33-bit designs rather than anything to do with the other job_descriptor_size/v4/v5 complication turned out to be telling. [ as an aside, are 64-bit jobs actually known not to work on v4 GPUs, or is it just that nobody's yet observed a 64-bit blob driving one? ] Long story short, it appears that 'Mali LPAE' is also lacking the start level notion of VMSA, and expects a full 4-level table even for <40 bits when level 0 effectively redundant. Thus walking the 3-level table that io-pgtable comes back with ends up going wildly wrong. The hack below seems to do the job for me; if Clément can confirm (on T-720 you'll still need the userspace hack to force 32-bit jobs as well) then I think I'll cook up a proper refactoring of the allocator to put things right. Robin. ----->8----- diff --git a/drivers/iommu/io-pgtable-arm.c b/drivers/iommu/io-pgtable-arm.c index 546968d8a349..f29da6e8dc08 100644 --- a/drivers/iommu/io-pgtable-arm.c +++ b/drivers/iommu/io-pgtable-arm.c @@ -1023,12 +1023,14 @@ arm_mali_lpae_alloc_pgtable(struct io_pgtable_cfg *cfg, void *cookie) iop = arm_64_lpae_alloc_pgtable_s1(cfg, cookie); if (iop) { u64 mair, ttbr; + struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(&iop->ops); + data->levels = 4; /* Copy values as union fields overlap */ mair = cfg->arm_lpae_s1_cfg.mair[0]; ttbr = cfg->arm_lpae_s1_cfg.ttbr[0];
On Wed, 29 May 2019 at 19:38, Robin Murphy <robin.murphy@arm.com> wrote: > > On 29/05/2019 16:09, Tomeu Vizoso wrote: > > On Tue, 21 May 2019 at 18:11, Clément Péron <peron.clem@gmail.com> wrote: > >> > > [snip] > >> [ 345.204813] panfrost 1800000.gpu: mmu irq status=1 > >> [ 345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > >> 0x0000000002400400 > > > > From what I can see here, 0x0000000002400400 points to the first byte > > of the first submitted job descriptor. > > > > So mapping buffers for the GPU doesn't seem to be working at all on > > 64-bit T-760. > > > > Steven, Robin, do you have any idea of why this could be? > > I tried rolling back to the old panfrost/nondrm shim, and it works fine > with kbase, and I also found that T-820 falls over in the exact same > manner, so the fact that it seemed to be common to the smaller 33-bit > designs rather than anything to do with the other > job_descriptor_size/v4/v5 complication turned out to be telling. Is this complication something you can explain? I don't know what v4 and v5 are meant here. > [ as an aside, are 64-bit jobs actually known not to work on v4 GPUs, or > is it just that nobody's yet observed a 64-bit blob driving one? ] I'm looking right now at getting Panfrost working on T720 with 64-bit descriptors, with the ultimate goal of making Panfrost 64-bit-descriptor only so we can have a single build of Mesa in distros. > Long story short, it appears that 'Mali LPAE' is also lacking the start > level notion of VMSA, and expects a full 4-level table even for <40 bits > when level 0 effectively redundant. Thus walking the 3-level table that > io-pgtable comes back with ends up going wildly wrong. The hack below > seems to do the job for me; if Clément can confirm (on T-720 you'll > still need the userspace hack to force 32-bit jobs as well) then I think > I'll cook up a proper refactoring of the allocator to put things right. Mmaps seem to work with this patch, thanks. The main complication I'm facing right now seems to be that the SFBD descriptor on T720 seems to be different from the one we already had (tested on T6xx?). Cheers, Tomeu > Robin. > > > ----->8----- > diff --git a/drivers/iommu/io-pgtable-arm.c b/drivers/iommu/io-pgtable-arm.c > index 546968d8a349..f29da6e8dc08 100644 > --- a/drivers/iommu/io-pgtable-arm.c > +++ b/drivers/iommu/io-pgtable-arm.c > @@ -1023,12 +1023,14 @@ arm_mali_lpae_alloc_pgtable(struct > io_pgtable_cfg *cfg, void *cookie) > iop = arm_64_lpae_alloc_pgtable_s1(cfg, cookie); > if (iop) { > u64 mair, ttbr; > + struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(&iop->ops); > > + data->levels = 4; > /* Copy values as union fields overlap */ > mair = cfg->arm_lpae_s1_cfg.mair[0]; > ttbr = cfg->arm_lpae_s1_cfg.ttbr[0]; > > _______________________________________________ > dri-devel mailing list > dri-devel@lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/dri-devel
On 31/05/2019 13:04, Tomeu Vizoso wrote: > On Wed, 29 May 2019 at 19:38, Robin Murphy <robin.murphy@arm.com> wrote: >> >> On 29/05/2019 16:09, Tomeu Vizoso wrote: >>> On Tue, 21 May 2019 at 18:11, Clément Péron <peron.clem@gmail.com> wrote: >>>> >>> [snip] >>>> [ 345.204813] panfrost 1800000.gpu: mmu irq status=1 >>>> [ 345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA >>>> 0x0000000002400400 >>> >>> From what I can see here, 0x0000000002400400 points to the first byte >>> of the first submitted job descriptor. >>> >>> So mapping buffers for the GPU doesn't seem to be working at all on >>> 64-bit T-760. >>> >>> Steven, Robin, do you have any idea of why this could be? >> >> I tried rolling back to the old panfrost/nondrm shim, and it works fine >> with kbase, and I also found that T-820 falls over in the exact same >> manner, so the fact that it seemed to be common to the smaller 33-bit >> designs rather than anything to do with the other >> job_descriptor_size/v4/v5 complication turned out to be telling. > > Is this complication something you can explain? I don't know what v4 > and v5 are meant here. I was alluding to BASE_HW_FEATURE_V4, which I believe refers to the Midgard architecture version - the older versions implemented by T6xx and T720 seem to be collectively treated as "v4", while T760 and T8xx would effectively be "v5". >> [ as an aside, are 64-bit jobs actually known not to work on v4 GPUs, or >> is it just that nobody's yet observed a 64-bit blob driving one? ] > > I'm looking right now at getting Panfrost working on T720 with 64-bit > descriptors, with the ultimate goal of making Panfrost > 64-bit-descriptor only so we can have a single build of Mesa in > distros. Cool, I'll keep an eye out, and hope that it might be enough for T620 on Juno, too :) >> Long story short, it appears that 'Mali LPAE' is also lacking the start >> level notion of VMSA, and expects a full 4-level table even for <40 bits >> when level 0 effectively redundant. Thus walking the 3-level table that >> io-pgtable comes back with ends up going wildly wrong. The hack below >> seems to do the job for me; if Clément can confirm (on T-720 you'll >> still need the userspace hack to force 32-bit jobs as well) then I think >> I'll cook up a proper refactoring of the allocator to put things right. > > Mmaps seem to work with this patch, thanks. > > The main complication I'm facing right now seems to be that the SFBD > descriptor on T720 seems to be different from the one we already had > (tested on T6xx?). OK - with the 32-bit hack pointed to up-thread, a quick kmscube test gave me the impression that T720 works fine, but on closer inspection some parts of glmark2 do seem to go a bit wonky (although I suspect at least some of it is just down to the FPGA setup being both very slow and lacking in memory bandwidth), and the "nv12-1img" mode of kmscube turns out to render in some delightfully wrong colours. I'll try to get a 'proper' version of the io-pgtable patch posted soon. Thanks, Robin. > > Cheers, > > Tomeu > >> Robin. >> >> >> ----->8----- >> diff --git a/drivers/iommu/io-pgtable-arm.c b/drivers/iommu/io-pgtable-arm.c >> index 546968d8a349..f29da6e8dc08 100644 >> --- a/drivers/iommu/io-pgtable-arm.c >> +++ b/drivers/iommu/io-pgtable-arm.c >> @@ -1023,12 +1023,14 @@ arm_mali_lpae_alloc_pgtable(struct >> io_pgtable_cfg *cfg, void *cookie) >> iop = arm_64_lpae_alloc_pgtable_s1(cfg, cookie); >> if (iop) { >> u64 mair, ttbr; >> + struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(&iop->ops); >> >> + data->levels = 4; >> /* Copy values as union fields overlap */ >> mair = cfg->arm_lpae_s1_cfg.mair[0]; >> ttbr = cfg->arm_lpae_s1_cfg.ttbr[0]; >> >> _______________________________________________ >> dri-devel mailing list >> dri-devel@lists.freedesktop.org >> https://lists.freedesktop.org/mailman/listinfo/dri-devel
Hi Maxime, Joerg, On Wed, 22 May 2019 at 21:27, Rob Herring <robh+dt@kernel.org> wrote: > > On Tue, May 21, 2019 at 11:11 AM Clément Péron <peron.clem@gmail.com> wrote: > > > > Hi, > > > > The Allwinner H6 has a Mali-T720 MP2 which should be supported by > > the new panfrost driver. This series fix two issues and introduce the > > dt-bindings but a simple benchmark show that it's still NOT WORKING. > > > > I'm pushing it in case someone want to continue the work. > > > > This has been tested with Mesa3D 19.1.0-RC2 and a GPU bitness patch[1]. > > > > One patch is from Icenowy Zheng where I changed the order as required > > by Rob Herring[2]. > > > > Thanks, > > Clement > > > > [1] https://gitlab.freedesktop.org/kszaq/mesa/tree/panfrost_64_32 > > [2] https://patchwork.kernel.org/patch/10699829/ > > > > > > [ 345.204813] panfrost 1800000.gpu: mmu irq status=1 > > [ 345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > > 0x0000000002400400 > > [ 345.209617] Reason: TODO > > [ 345.209617] raw fault status: 0x800002C1 > > [ 345.209617] decoded fault status: SLAVE FAULT > > [ 345.209617] exception type 0xC1: TRANSLATION_FAULT_LEVEL1 > > [ 345.209617] access type 0x2: READ > > [ 345.209617] source id 0x8000 > > [ 345.729957] panfrost 1800000.gpu: gpu sched timeout, js=0, > > status=0x8, head=0x2400400, tail=0x2400400, sched_job=000000009e204de9 > > [ 346.055876] panfrost 1800000.gpu: mmu irq status=1 > > [ 346.060680] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > > 0x0000000002C00A00 > > [ 346.060680] Reason: TODO > > [ 346.060680] raw fault status: 0x810002C1 > > [ 346.060680] decoded fault status: SLAVE FAULT > > [ 346.060680] exception type 0xC1: TRANSLATION_FAULT_LEVEL1 > > [ 346.060680] access type 0x2: READ > > [ 346.060680] source id 0x8100 > > [ 346.561955] panfrost 1800000.gpu: gpu sched timeout, js=1, > > status=0x8, head=0x2c00a00, tail=0x2c00a00, sched_job=00000000b55a9a85 > > [ 346.573913] panfrost 1800000.gpu: mmu irq status=1 > > [ 346.578707] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > > 0x0000000002C00B80 > > > > Change in v5: > > - Remove fix indent > > > > Changes in v4: > > - Add bus_clock probe > > - Fix sanity check in io-pgtable > > - Add vramp-delay > > - Merge all boards into one patch > > - Remove upstreamed Neil A. patch > > > > Change in v3 (Thanks to Maxime Ripard): > > - Reauthor Icenowy for her path > > > > Changes in v2 (Thanks to Maxime Ripard): > > - Drop GPU OPP Table > > - Add clocks and clock-names in required > > > > Clément Péron (5): > > drm: panfrost: add optional bus_clock > > iommu: io-pgtable: fix sanity check for non 48-bit mali iommu > > dt-bindings: gpu: mali-midgard: Add H6 mali gpu compatible > > arm64: dts: allwinner: Add ARM Mali GPU node for H6 > > arm64: dts: allwinner: Add mali GPU supply for H6 boards > > > > Icenowy Zheng (1): > > dt-bindings: gpu: add bus clock for Mali Midgard GPUs > > I've applied patches 1 and 3 to drm-misc. I was going to do patch 4 > too, but it doesn't apply. > > Patch 2 can go in via the iommu tree and the rest via the allwinner tree. Is this OK for you to pick up this series? Thanks, Clément > > Rob
On Wed, 29 May 2019 at 19:38, Robin Murphy <robin.murphy@arm.com> wrote: > > On 29/05/2019 16:09, Tomeu Vizoso wrote: > > On Tue, 21 May 2019 at 18:11, Clément Péron <peron.clem@gmail.com> wrote: > >> > > [snip] > >> [ 345.204813] panfrost 1800000.gpu: mmu irq status=1 > >> [ 345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA > >> 0x0000000002400400 > > > > From what I can see here, 0x0000000002400400 points to the first byte > > of the first submitted job descriptor. > > > > So mapping buffers for the GPU doesn't seem to be working at all on > > 64-bit T-760. > > > > Steven, Robin, do you have any idea of why this could be? > > I tried rolling back to the old panfrost/nondrm shim, and it works fine > with kbase, and I also found that T-820 falls over in the exact same > manner, so the fact that it seemed to be common to the smaller 33-bit > designs rather than anything to do with the other > job_descriptor_size/v4/v5 complication turned out to be telling. > > [ as an aside, are 64-bit jobs actually known not to work on v4 GPUs, or > is it just that nobody's yet observed a 64-bit blob driving one? ] Do you know if 64-bit descriptors work on v4 GPUs with our kernel driver but with the DDK? Wonder if there something else to be fixed in the kernel for that scenario. Thanks, Tomeu > Long story short, it appears that 'Mali LPAE' is also lacking the start > level notion of VMSA, and expects a full 4-level table even for <40 bits > when level 0 effectively redundant. Thus walking the 3-level table that > io-pgtable comes back with ends up going wildly wrong. The hack below > seems to do the job for me; if Clément can confirm (on T-720 you'll > still need the userspace hack to force 32-bit jobs as well) then I think > I'll cook up a proper refactoring of the allocator to put things right. > > Robin. > > > ----->8----- > diff --git a/drivers/iommu/io-pgtable-arm.c b/drivers/iommu/io-pgtable-arm.c > index 546968d8a349..f29da6e8dc08 100644 > --- a/drivers/iommu/io-pgtable-arm.c > +++ b/drivers/iommu/io-pgtable-arm.c > @@ -1023,12 +1023,14 @@ arm_mali_lpae_alloc_pgtable(struct > io_pgtable_cfg *cfg, void *cookie) > iop = arm_64_lpae_alloc_pgtable_s1(cfg, cookie); > if (iop) { > u64 mair, ttbr; > + struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(&iop->ops); > > + data->levels = 4; > /* Copy values as union fields overlap */ > mair = cfg->arm_lpae_s1_cfg.mair[0]; > ttbr = cfg->arm_lpae_s1_cfg.ttbr[0]; > > _______________________________________________ > dri-devel mailing list > dri-devel@lists.freedesktop.org > https://lists.freedesktop.org/mailman/listinfo/dri-devel
Hi Robyn, On 31/05/2019 15:47, Robin Murphy wrote: > On 31/05/2019 13:04, Tomeu Vizoso wrote: >> On Wed, 29 May 2019 at 19:38, Robin Murphy <robin.murphy@arm.com> wrote: >>> >>> On 29/05/2019 16:09, Tomeu Vizoso wrote: >>>> On Tue, 21 May 2019 at 18:11, Clément Péron <peron.clem@gmail.com> wrote: >>>>> >>>> [snip] >>>>> [ 345.204813] panfrost 1800000.gpu: mmu irq status=1 >>>>> [ 345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA >>>>> 0x0000000002400400 >>>> >>>> From what I can see here, 0x0000000002400400 points to the first byte >>>> of the first submitted job descriptor. >>>> >>>> So mapping buffers for the GPU doesn't seem to be working at all on >>>> 64-bit T-760. >>>> >>>> Steven, Robin, do you have any idea of why this could be? >>> >>> I tried rolling back to the old panfrost/nondrm shim, and it works fine >>> with kbase, and I also found that T-820 falls over in the exact same >>> manner, so the fact that it seemed to be common to the smaller 33-bit >>> designs rather than anything to do with the other >>> job_descriptor_size/v4/v5 complication turned out to be telling. >> >> Is this complication something you can explain? I don't know what v4 >> and v5 are meant here. > > I was alluding to BASE_HW_FEATURE_V4, which I believe refers to the Midgard architecture version - the older versions implemented by T6xx and T720 seem to be collectively treated as "v4", while T760 and T8xx would effectively be "v5". > >>> [ as an aside, are 64-bit jobs actually known not to work on v4 GPUs, or >>> is it just that nobody's yet observed a 64-bit blob driving one? ] >> >> I'm looking right now at getting Panfrost working on T720 with 64-bit >> descriptors, with the ultimate goal of making Panfrost >> 64-bit-descriptor only so we can have a single build of Mesa in >> distros. > > Cool, I'll keep an eye out, and hope that it might be enough for T620 on Juno, too :) > >>> Long story short, it appears that 'Mali LPAE' is also lacking the start >>> level notion of VMSA, and expects a full 4-level table even for <40 bits >>> when level 0 effectively redundant. Thus walking the 3-level table that >>> io-pgtable comes back with ends up going wildly wrong. The hack below >>> seems to do the job for me; if Clément can confirm (on T-720 you'll >>> still need the userspace hack to force 32-bit jobs as well) then I think >>> I'll cook up a proper refactoring of the allocator to put things right. >> >> Mmaps seem to work with this patch, thanks. >> >> The main complication I'm facing right now seems to be that the SFBD >> descriptor on T720 seems to be different from the one we already had >> (tested on T6xx?). > > OK - with the 32-bit hack pointed to up-thread, a quick kmscube test gave me the impression that T720 works fine, but on closer inspection some parts of glmark2 do seem to go a bit wonky (although I suspect at least some of it is just down to the FPGA setup being both very slow and lacking in memory bandwidth), and the "nv12-1img" mode of kmscube turns out to render in some delightfully wrong colours. > > I'll try to get a 'proper' version of the io-pgtable patch posted soon. I'm trying to collect all the fixes needed to make T820 work again, and I was wondering if you finally have a proper patch for this and "cfg->ias > 48" hack ? Or one I can test ? Thanks, Neil > > Thanks, > Robin. > >> >> Cheers, >> >> Tomeu >> >>> Robin. >>> >>> >>> ----->8----- >>> diff --git a/drivers/iommu/io-pgtable-arm.c b/drivers/iommu/io-pgtable-arm.c >>> index 546968d8a349..f29da6e8dc08 100644 >>> --- a/drivers/iommu/io-pgtable-arm.c >>> +++ b/drivers/iommu/io-pgtable-arm.c >>> @@ -1023,12 +1023,14 @@ arm_mali_lpae_alloc_pgtable(struct >>> io_pgtable_cfg *cfg, void *cookie) >>> iop = arm_64_lpae_alloc_pgtable_s1(cfg, cookie); >>> if (iop) { >>> u64 mair, ttbr; >>> + struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(&iop->ops); >>> >>> + data->levels = 4; >>> /* Copy values as union fields overlap */ >>> mair = cfg->arm_lpae_s1_cfg.mair[0]; >>> ttbr = cfg->arm_lpae_s1_cfg.ttbr[0]; >>> >>> _______________________________________________ >>> dri-devel mailing list >>> dri-devel@lists.freedesktop.org >>> https://lists.freedesktop.org/mailman/listinfo/dri-devel
Hi Neil, On 28/08/2019 12:28, Neil Armstrong wrote: > Hi Robin, > > On 31/05/2019 15:47, Robin Murphy wrote: >> On 31/05/2019 13:04, Tomeu Vizoso wrote: >>> On Wed, 29 May 2019 at 19:38, Robin Murphy <robin.murphy@arm.com> wrote: >>>> >>>> On 29/05/2019 16:09, Tomeu Vizoso wrote: >>>>> On Tue, 21 May 2019 at 18:11, Clément Péron <peron.clem@gmail.com> wrote: >>>>>> >>>>> [snip] >>>>>> [ 345.204813] panfrost 1800000.gpu: mmu irq status=1 >>>>>> [ 345.209617] panfrost 1800000.gpu: Unhandled Page fault in AS0 at VA >>>>>> 0x0000000002400400 >>>>> >>>>> From what I can see here, 0x0000000002400400 points to the first byte >>>>> of the first submitted job descriptor. >>>>> >>>>> So mapping buffers for the GPU doesn't seem to be working at all on >>>>> 64-bit T-760. >>>>> >>>>> Steven, Robin, do you have any idea of why this could be? >>>> >>>> I tried rolling back to the old panfrost/nondrm shim, and it works fine >>>> with kbase, and I also found that T-820 falls over in the exact same >>>> manner, so the fact that it seemed to be common to the smaller 33-bit >>>> designs rather than anything to do with the other >>>> job_descriptor_size/v4/v5 complication turned out to be telling. >>> >>> Is this complication something you can explain? I don't know what v4 >>> and v5 are meant here. >> >> I was alluding to BASE_HW_FEATURE_V4, which I believe refers to the Midgard architecture version - the older versions implemented by T6xx and T720 seem to be collectively treated as "v4", while T760 and T8xx would effectively be "v5". >> >>>> [ as an aside, are 64-bit jobs actually known not to work on v4 GPUs, or >>>> is it just that nobody's yet observed a 64-bit blob driving one? ] >>> >>> I'm looking right now at getting Panfrost working on T720 with 64-bit >>> descriptors, with the ultimate goal of making Panfrost >>> 64-bit-descriptor only so we can have a single build of Mesa in >>> distros. >> >> Cool, I'll keep an eye out, and hope that it might be enough for T620 on Juno, too :) >> >>>> Long story short, it appears that 'Mali LPAE' is also lacking the start >>>> level notion of VMSA, and expects a full 4-level table even for <40 bits >>>> when level 0 effectively redundant. Thus walking the 3-level table that >>>> io-pgtable comes back with ends up going wildly wrong. The hack below >>>> seems to do the job for me; if Clément can confirm (on T-720 you'll >>>> still need the userspace hack to force 32-bit jobs as well) then I think >>>> I'll cook up a proper refactoring of the allocator to put things right. >>> >>> Mmaps seem to work with this patch, thanks. >>> >>> The main complication I'm facing right now seems to be that the SFBD >>> descriptor on T720 seems to be different from the one we already had >>> (tested on T6xx?). >> >> OK - with the 32-bit hack pointed to up-thread, a quick kmscube test gave me the impression that T720 works fine, but on closer inspection some parts of glmark2 do seem to go a bit wonky (although I suspect at least some of it is just down to the FPGA setup being both very slow and lacking in memory bandwidth), and the "nv12-1img" mode of kmscube turns out to render in some delightfully wrong colours. >> >> I'll try to get a 'proper' version of the io-pgtable patch posted soon. > > I'm trying to collect all the fixes needed to make T820 work again, and > I was wondering if you finally have a proper patch for this and "cfg->ias > 48" > hack ? Or one I can test ? I do have a handful of io-pgtable patches written up and ready to go, I'm just treading carefully and waiting for the internal approval box to be ticked before I share anything :( Robin. > > Thanks, > Neil > >> >> Thanks, >> Robin. >> >>> >>> Cheers, >>> >>> Tomeu >>> >>>> Robin. >>>> >>>> >>>> ----->8----- >>>> diff --git a/drivers/iommu/io-pgtable-arm.c b/drivers/iommu/io-pgtable-arm.c >>>> index 546968d8a349..f29da6e8dc08 100644 >>>> --- a/drivers/iommu/io-pgtable-arm.c >>>> +++ b/drivers/iommu/io-pgtable-arm.c >>>> @@ -1023,12 +1023,14 @@ arm_mali_lpae_alloc_pgtable(struct >>>> io_pgtable_cfg *cfg, void *cookie) >>>> iop = arm_64_lpae_alloc_pgtable_s1(cfg, cookie); >>>> if (iop) { >>>> u64 mair, ttbr; >>>> + struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(&iop->ops); >>>> >>>> + data->levels = 4; >>>> /* Copy values as union fields overlap */ >>>> mair = cfg->arm_lpae_s1_cfg.mair[0]; >>>> ttbr = cfg->arm_lpae_s1_cfg.ttbr[0]; >>>> >>>> _______________________________________________ >>>> dri-devel mailing list >>>> dri-devel@lists.freedesktop.org >>>> https://lists.freedesktop.org/mailman/listinfo/dri-devel >
On 28/08/2019 13:49, Robin Murphy wrote: > Hi Neil, > > On 28/08/2019 12:28, Neil Armstrong wrote: >> Hi Robin, >> [...] >>> >>> OK - with the 32-bit hack pointed to up-thread, a quick kmscube test gave me the impression that T720 works fine, but on closer inspection some parts of glmark2 do seem to go a bit wonky (although I suspect at least some of it is just down to the FPGA setup being both very slow and lacking in memory bandwidth), and the "nv12-1img" mode of kmscube turns out to render in some delightfully wrong colours. >>> >>> I'll try to get a 'proper' version of the io-pgtable patch posted soon. >> >> I'm trying to collect all the fixes needed to make T820 work again, and >> I was wondering if you finally have a proper patch for this and "cfg->ias > 48" >> hack ? Or one I can test ? > > I do have a handful of io-pgtable patches written up and ready to go, I'm just treading carefully and waiting for the internal approval box to be ticked before I share anything :( Great ! No problem, it can totally wait until approval, Thanks, Neil > > Robin. > >> >> Thanks, >> Neil >> >>> >>> Thanks, >>> Robin. >>> >>>> >>>> Cheers, >>>> >>>> Tomeu >>>> >>>>> Robin. >>>>> >>>>> >>>>> ----->8----- >>>>> diff --git a/drivers/iommu/io-pgtable-arm.c b/drivers/iommu/io-pgtable-arm.c >>>>> index 546968d8a349..f29da6e8dc08 100644 >>>>> --- a/drivers/iommu/io-pgtable-arm.c >>>>> +++ b/drivers/iommu/io-pgtable-arm.c >>>>> @@ -1023,12 +1023,14 @@ arm_mali_lpae_alloc_pgtable(struct >>>>> io_pgtable_cfg *cfg, void *cookie) >>>>> iop = arm_64_lpae_alloc_pgtable_s1(cfg, cookie); >>>>> if (iop) { >>>>> u64 mair, ttbr; >>>>> + struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(&iop->ops); >>>>> >>>>> + data->levels = 4; >>>>> /* Copy values as union fields overlap */ >>>>> mair = cfg->arm_lpae_s1_cfg.mair[0]; >>>>> ttbr = cfg->arm_lpae_s1_cfg.ttbr[0]; >>>>> >>>>> _______________________________________________ >>>>> dri-devel mailing list >>>>> dri-devel@lists.freedesktop.org >>>>> https://lists.freedesktop.org/mailman/listinfo/dri-devel >>