drm: add func to better detect wether swiotlb is needed
Subsystems (3)
PATCH 1/1 drm: add func to better detect wether swiotlb is needed
This commit fixes DRM failures on Xen PV systems that were introduced in v4.17 by the following commits:
82626363 drm: add func to get max iomem address v2 fd5fd480 drm/amdgpu: only enable swiotlb alloc when need v2 1bc3d3cc drm/radeon: only enable swiotlb path when need v2
The introduction of ->need_swiotlb to the ttm_dma_populate() conditionals in the radeon and amdgpu device drivers causes Gnome to immediately crash on Xen PV systems, returning the user to the login screen. The following kernel errors get logged:
[ 28.554259] radeon_dp_aux_transfer_native: 200 callbacks suppressed [ 31.219821] radeon 0000:01:00.0: swiotlb buffer is full (sz: 2097152 bytes) [ 31.220030] [drm:radeon_gem_object_create [radeon]] *ERROR* Failed to allocate GEM object (16384000, 2, 4096, -14) [ 31.226109] radeon 0000:01:00.0: swiotlb buffer is full (sz: 2097152 bytes) [ 31.226300] [drm:radeon_gem_object_create [radeon]] *ERROR* Failed to allocate GEM object (16384000, 2, 4096, -14) [ 31.300734] gnome-shell[1935]: segfault at 88 ip 00007f39151cd904 sp 00007ffc97611ad8 error 4 in libmutter-cogl.so[7f3915178000+aa000] [ 31.300745] Code: 5f c3 0f 1f 40 00 48 8b 47 78 48 8b 40 40 ff e0 66 0f 1f 44 00 00 48 8b 47 78 48 8b 40 48 ff e0 66 0f 1f 44 00 00 48 8b 47 78 <48> 8b 80 88 00 00 00 ff e0 0f 1f 00 48 8b 47 78 48 8b 40 68 ff e0 [ 38.193302] radeon_dp_aux_transfer_native: 116 callbacks suppressed [ 40.009317] radeon 0000:01:00.0: swiotlb buffer is full (sz: 2097152 bytes) [ 40.009488] [drm:radeon_gem_object_create [radeon]] *ERROR* Failed to allocate GEM object (16384000, 2, 4096, -14) [ 40.015114] radeon 0000:01:00.0: swiotlb buffer is full (sz: 2097152 bytes) [ 40.015297] [drm:radeon_gem_object_create [radeon]] *ERROR* Failed to allocate GEM object (16384000, 2, 4096, -14) [ 40.028302] gnome-shell[2431]: segfault at 2dadf40 ip 0000000002dadf40 sp 00007ffcd24ea5f8 error 15 [ 40.028306] Code: 20 6e 31 00 00 00 00 00 00 00 00 37 e3 3d 2d 7f 00 00 80 f4 e6 3d 2d 7f 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 <00> 00 00 00 00 00 00 00 c1 00 00 00 00 00 00 00 80 e1 d2 03 00 00
This commit renames drm_get_max_iomem() to drm_need_swiotlb(), adds a xen_pv_domain() check to it, and moves the bit shifting comparison that always follows its usage into the function (simplifying the drm driver code).
--- drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c | 2 +- drivers/gpu/drm/drm_memory.c | 19 ++++++++++++++++--- drivers/gpu/drm/radeon/radeon_device.c | 2 +- include/drm/drm_cache.h | 2 +- 6 files changed, 21 insertions(+), 8 deletions(-)
drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c +1 -1
@@ -1029,7 +1029,7 @@ static int gmc_v7_0_sw_init(void *handle)pci_set_consistent_dma_mask (adev->pdev,DMA_BIT_MASK (32 ));pr_warn ("amdgpu: No coherent DMA available\n" ); } - adev->need_swiotlb =drm_get_max_iomem () > ((u64 )1 << dma_bits); + adev->need_swiotlb =drm_need_swiotlb (dma_bits); r =gmc_v7_0_init_microcode (adev);if (r) {
drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c +1 -1
@@ -1155,7 +1155,7 @@ static int gmc_v8_0_sw_init(void *handle)pci_set_consistent_dma_mask (adev->pdev,DMA_BIT_MASK (32 ));pr_warn ("amdgpu: No coherent DMA available\n" ); } - adev->need_swiotlb =drm_get_max_iomem () > ((u64 )1 << dma_bits); + adev->need_swiotlb =drm_need_swiotlb (dma_bits); r =gmc_v8_0_init_microcode (adev);if (r) {
drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +1 -1
@@ -989,7 +989,7 @@ static int gmc_v9_0_sw_init(void *handle)pci_set_consistent_dma_mask (adev->pdev,DMA_BIT_MASK (32 ));printk (KERN_WARNING "amdgpu: No coherent DMA available.\n" ); } - adev->need_swiotlb =drm_get_max_iomem () > ((u64 )1 << dma_bits); + adev->need_swiotlb =drm_need_swiotlb (dma_bits);if (adev->asic_type ==CHIP_VEGA20 ) { r =gfxhub_v1_1_get_xgmi_info (adev);
drivers/gpu/drm/drm_memory.c +16 -3
@@ -35,6 +35,7 @@#include <linux/highmem.h> #include <linux/export.h> +#include <xen/xen.h> #include <drm/drmP.h> #include "drm_legacy.h"
@@ -150,15 +151,27 @@ void drm_legacy_ioremapfree(struct drm_local_map *map, struct drm_device *dev) }EXPORT_SYMBOL (drm_legacy_ioremapfree); -u64 drm_get_max_iomem (void ) +bool drm_need_swiotlb (int dma_bits) {struct resource *tmp;resource_size_t max_iomem =0 ; +/* +* Xen paravirtual hosts require swiotlb regardless of requested dma +* transfer size. +* +* NOTE: Really, what it requires is use of the dma_alloc_coherent +* allocator used in ttm_dma_populate() instead of +* ttm_populate_and_map_pages(), which bounce buffers so much in +* Xen it leads to swiotlb buffer exhaustion. +*/ +if (xen_pv_domain ()) +return true; +for (tmp = iomem_resource.child; tmp; tmp = tmp->sibling) { max_iomem =max (max_iomem, tmp->end); } -return max_iomem; +return max_iomem > ((u64 )1 << dma_bits); } -EXPORT_SYMBOL (drm_get_max_iomem); +EXPORT_SYMBOL (drm_need_swiotlb);
drivers/gpu/drm/radeon/radeon_device.c +1 -1
@@ -1387,7 +1387,7 @@ int radeon_device_init(struct radeon_device *rdev,pci_set_consistent_dma_mask (rdev->pdev,DMA_BIT_MASK (32 ));pr_warn ("radeon: No coherent DMA available\n" ); } - rdev->need_swiotlb =drm_get_max_iomem () > ((u64 )1 << dma_bits); + rdev->need_swiotlb =drm_need_swiotlb (dma_bits);/* Registers mapping */ /* TODO: block userspace mapping of io register */
include/drm/drm_cache.h +1 -1
@@ -38,7 +38,7 @@void drm_clflush_pages (struct page *pages[],unsigned long num_pages);void drm_clflush_sg (struct sg_table *st);void drm_clflush_virt_range (void *addr,unsigned long length); -u64 drm_get_max_iomem (void ); +bool drm_need_swiotlb (int dma_bits);static inline bool drm_arch_can_wc_memory (void )
Signature · 7 lines
1.8.3.1 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Quoted 143 lines
-----Original Message-----
From: Michael D Labriola [mailto:[email protected]]
Sent: 19 February 2019 23:08
To: [email protected]; Alex Deucher <[email protected]>; Christian Koenig <[email protected]>; Chunming Zhou <[email protected]>; [email protected]; Monk Liu <[email protected]>
Cc: Juergen Gross <[email protected]>; Christoph Hellwig <[email protected]>; Andrew Cooper <[email protected]>; Paul Durrant <[email protected]>; [email protected]; Jan Beulich <[email protected]>; Konrad Rzeszutek Wilk <[email protected]>; Michael D Labriola <[email protected]>
Subject: [PATCH] drm: add func to better detect wether swiotlb is neededThis commit fixes DRM failures on Xen PV systems that were introduced in v4.17 by the following commits:
82626363 drm: add func to get max iomem address v2
fd5fd480drm/amdgpu: only enable swiotlb alloc when need v21bc3d3ccdrm/radeon: only enable swiotlb path when need v2The introduction of ->need_swiotlb to the
ttm_dma_populate()conditionals in the radeon and amdgpu device drivers causes Gnome to immediately crash on Xen PV systems, returning the user to the login screen. The following kernel errors get logged:[ 28.554259] radeon_dp_aux_transfer_native: 200 callbacks suppressed [ 31.219821] radeon 0000:01:00.0: swiotlb buffer is full (sz: 2097152 bytes) [ 31.220030] [drm:radeon_gem_object_create [radeon]] *ERROR* Failed to allocate GEM object (16384000, 2, 4096, -14) [ 31.226109] radeon 0000:01:00.0: swiotlb buffer is full (sz: 2097152 bytes) [ 31.226300] [drm:radeon_gem_object_create [radeon]] *ERROR* Failed to allocate GEM object (16384000, 2, 4096, -14) [ 31.300734] gnome-shell[1935]: segfault at 88 ip 00007f39151cd904 sp 00007ffc97611ad8 error 4 in libmutter-cogl.so[7f3915178000+aa000] [ 31.300745] Code: 5f c3 0f 1f 40 00 48 8b 47 78 48 8b 40 40 ff e0 66 0f 1f 44 00 00 48 8b 47 78 48 8b 40 48 ff e0 66 0f 1f 44 00 00 48 8b 47 78 <48> 8b 80 88 00 00 00 ff e0 0f 1f 00 48 8b 47 78 48 8b 40 68 ff e0 [ 38.193302] radeon_dp_aux_transfer_native: 116 callbacks suppressed [ 40.009317] radeon 0000:01:00.0: swiotlb buffer is full (sz: 2097152 bytes) [ 40.009488] [drm:radeon_gem_object_create [radeon]] *ERROR* Failed to allocate GEM object (16384000, 2, 4096, -14) [ 40.015114] radeon 0000:01:00.0: swiotlb buffer is full (sz: 2097152 bytes) [ 40.015297] [drm:radeon_gem_object_create [radeon]] *ERROR* Failed to allocate GEM object (16384000, 2, 4096, -14) [ 40.028302] gnome-shell[2431]: segfault at 2dadf40 ip 0000000002dadf40 sp 00007ffcd24ea5f8 error 15 [ 40.028306] Code: 20 6e 31 00 00 00 00 00 00 00 00 37 e3 3d 2d 7f 00 0080 f4 e6 3d 2d 7f 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 <00> 00 00 00 00 00 00 00 c1 00 00 00 00 00 00 00 80 e1 d2 03 00 00
This commit renames
drm_get_max_iomem()todrm_need_swiotlb(), adds axen_pv_domain()check to it, and moves the bit shifting comparison that always follows its usage into the function (simplifying the drm driver code).Signed-off-by Michael D Labriola <[email protected]>--- drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c | 2 +- drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c | 2 +- drivers/gpu/drm/drm_memory.c | 19 ++++++++++++++++--- drivers/gpu/drm/radeon/radeon_device.c | 2 +- include/drm/drm_cache.h | 2 +- 6 files changed, 21 insertions(+), 8 deletions(-)drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c +1
-1@@ -1029,7 +1029,7 @@ static int gmc_v7_0_sw_init(void *handle)pci_set_consistent_dma_mask (adev->pdev,DMA_BIT_MASK (32 ));pr_warn ("amdgpu: No coherent DMA available\n" ); } - adev->need_swiotlb =drm_get_max_iomem () > ((u64 )1 << dma_bits); + adev->need_swiotlb =drm_need_swiotlb (dma_bits); r =gmc_v7_0_init_microcode (adev);if (r) {drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c +1
-1@@ -1155,7 +1155,7 @@ static int gmc_v8_0_sw_init(void *handle)pci_set_consistent_dma_mask (adev->pdev,DMA_BIT_MASK (32 ));pr_warn ("amdgpu: No coherent DMA available\n" ); } - adev->need_swiotlb =drm_get_max_iomem () > ((u64 )1 << dma_bits); + adev->need_swiotlb =drm_need_swiotlb (dma_bits); r =gmc_v8_0_init_microcode (adev);if (r) {drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +1
-1@@ -989,7 +989,7 @@ static int gmc_v9_0_sw_init(void *handle)pci_set_consistent_dma_mask (adev->pdev,DMA_BIT_MASK (32 ));printk (KERN_WARNING "amdgpu: No coherent DMA available.\n" ); } - adev->need_swiotlb =drm_get_max_iomem () > ((u64 )1 << dma_bits); + adev->need_swiotlb =drm_need_swiotlb (dma_bits);if (adev->asic_type ==CHIP_VEGA20 ) { r =gfxhub_v1_1_get_xgmi_info (adev);drivers/gpu/drm/drm_memory.c +9
-1@@ -35,6 +35,7 @@#include <linux/highmem.h> #include <linux/export.h> +#include <xen/xen.h> #include <drm/drmP.h> #include "drm_legacy.h" @@ -150,15 +151,27 @@ void drm_legacy_ioremapfree(struct drm_local_map *map, struct drm_device *dev) }EXPORT_SYMBOL (drm_legacy_ioremapfree); -u64 drm_get_max_iomem (void ) +bool drm_need_swiotlb (int dma_bits) {struct resource *tmp;resource_size_t max_iomem =0 ; +/* +* Xen paravirtual hosts require swiotlb regardless of requested dma +* transfer size. +* +* NOTE: Really, what it requires is use of the dma_alloc_coherent +* allocator used in ttm_dma_populate() instead of +* ttm_populate_and_map_pages(), which bounce buffers so much in
drivers/gpu/drm/drm_memory.c void drm_legacy_ioremapfree(struct drm_local_map *map, struct drm_device *dev) · line 168
Quoted 3 lines
drivers/gpu/drm/drm_memory.c +3
-0+* Xen it leads to swiotlb buffer exhaustion. +*/ +if (xen_pv_domain ())
I've not been following all of the ins and outs of the discussion on this so apologies if I'm missing some context, but...
This looks like the wrong test to me. I think it should be:
if ( xen_swiotlb )
drivers/gpu/drm/drm_memory.c void drm_legacy_ioremapfree(struct drm_local_map *map, struct drm_device *dev) · line 170
Quoted 2 lines
drivers/gpu/drm/drm_memory.c +2
-0+return true; +
Paul
Quoted 37 lines
drivers/gpu/drm/drm_memory.c +2
-2for (tmp = iomem_resource.child; tmp; tmp = tmp->sibling) { max_iomem =max (max_iomem, tmp->end); } -return max_iomem; +return max_iomem > ((u64 )1 << dma_bits); } -EXPORT_SYMBOL (drm_get_max_iomem); +EXPORT_SYMBOL (drm_need_swiotlb);drivers/gpu/drm/radeon/radeon_device.c +1
-1@@ -1387,7 +1387,7 @@ int radeon_device_init(struct radeon_device *rdev,pci_set_consistent_dma_mask (rdev->pdev,DMA_BIT_MASK (32 ));pr_warn ("radeon: No coherent DMA available\n" ); } - rdev->need_swiotlb =drm_get_max_iomem () > ((u64 )1 << dma_bits); + rdev->need_swiotlb =drm_need_swiotlb (dma_bits);/* Registers mapping */ /* TODO: block userspace mapping of io register */ include/drm/drm_cache.h +1
-2@@ -38,7 +38,7 @@void drm_clflush_pages (struct page *pages[],unsigned long num_pages);void drm_clflush_sg (struct sg_table *st);void drm_clflush_virt_range (void *addr,unsigned long length); -u64 drm_get_max_iomem (void ); +bool drm_need_swiotlb (int dma_bits);static inline bool drm_arch_can_wc_memory (void ) --
Quoted 1 line
1.8.3.1
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
.snip..
Quoted 20 lines
Quoted 14 lines
drivers/gpu/drm/drm_memory.c +8
-1-u64 drm_get_max_iomem (void ) +bool drm_need_swiotlb (int dma_bits) {struct resource *tmp;resource_size_t max_iomem =0 ; +/* +* Xen paravirtual hosts require swiotlb regardless of requested dma +* transfer size. +* +* NOTE: Really, what it requires is use of the dma_alloc_coherent +* allocator used in ttm_dma_populate() instead of +* ttm_populate_and_map_pages(), which bounce buffers so much in
drivers/gpu/drm/drm_memory.c void drm_legacy_ioremapfree(struct drm_local_map *map, struct drm_device *dev) · line 168
Quoted 3 lines
drivers/gpu/drm/drm_memory.c +3
-0+* Xen it leads to swiotlb buffer exhaustion. +*/ +if (xen_pv_domain ())I've not been following all of the ins and outs of the discussion on this so apologies if I'm missing some context, but...
Quoted 3 lines
This looks like the wrong test to me. I think it should be:
if (
xen_swiotlb)
Ah, that could be as well.
Quoted 1 line
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
.snip..
Quoted 20 lines
Quoted 14 lines
drivers/gpu/drm/drm_memory.c +8
-1-u64 drm_get_max_iomem (void ) +bool drm_need_swiotlb (int dma_bits) {struct resource *tmp;resource_size_t max_iomem =0 ; +/* +* Xen paravirtual hosts require swiotlb regardless of requested dma +* transfer size. +* +* NOTE: Really, what it requires is use of the dma_alloc_coherent +* allocator used in ttm_dma_populate() instead of +* ttm_populate_and_map_pages(), which bounce buffers so much in
drivers/gpu/drm/drm_memory.c void drm_legacy_ioremapfree(struct drm_local_map *map, struct drm_device *dev) · line 168
Quoted 3 lines
drivers/gpu/drm/drm_memory.c +3
-0+* Xen it leads to swiotlb buffer exhaustion. +*/ +if (xen_pv_domain ())I've not been following all of the ins and outs of the discussion on this so apologies if I'm missing some context, but...
Quoted 3 lines
This looks like the wrong test to me. I think it should be:
if (
xen_swiotlb)
Ah, that could be as well.
Quoted 1 line
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel