drm/radeon: have the callers of set_memory_*() check the return value
Subsystems (2)
cover drm/radeon: have the callers of set_memory_*() check the return value
Right now several architectures allow their set_memory_*() family of functions to fail, but callers may not be checking the return values. If set_memory_*() returns with an error, call-site assumptions may be infact wrong to assume that it would either succeed or not succeed at all. Ideally, the failure of set_memory_*() should be passed up the call stack, and callers should examine the failure and deal with it.
Need to fix the callers and add the __must_check attribute. They also may not provide any level of atomicity, in the sense that the memory protections may be left incomplete on failure. This issue likely has a few steps on effects architectures:
1)Have all callers of set_memory_*() helpers check the return value. 2)Add __must_check to all set_memory_*() helpers so that new uses do not ignore the return value.
3)Add atomicity to the calls so that the memory protections aren't left in a partial state.
This series is part of step 1. Make drm/radeon check the return value of set_memory_*().
Tianlin Li (2):
drm/radeon: have the callers of set_memory_*() check the return value drm/radeon: change call sites to handle return value properly.
drivers/gpu/drm/radeon/r100.c | 3 ++- drivers/gpu/drm/radeon/radeon.h | 2 +- drivers/gpu/drm/radeon/radeon_gart.c | 22 ++++++++++++++++++---- drivers/gpu/drm/radeon/rs400.c | 3 ++- 4 files changed, 23 insertions(+), 7 deletions(-)
Signature · 1 line
2.17.1
PATCH 1/2 drm/radeon: have the callers of set_memory_*() check the return value
Have the callers of set_memory_*() in drm/radeon check the return value. Change the return type of the callers properly.
--- drivers/gpu/drm/radeon/radeon.h | 2 +- drivers/gpu/drm/radeon/radeon_gart.c | 22 ++++++++++++++++++---- 2 files changed, 19 insertions(+), 5 deletions(-)
drivers/gpu/drm/radeon/radeon.h +1 -1
@@ -661,7 +661,7 @@ struct radeon_gart { };int radeon_gart_table_ram_alloc (struct radeon_device *rdev); -void radeon_gart_table_ram_free (struct radeon_device *rdev); +int radeon_gart_table_ram_free (struct radeon_device *rdev);int radeon_gart_table_vram_alloc (struct radeon_device *rdev);void radeon_gart_table_vram_free (struct radeon_device *rdev);int radeon_gart_table_vram_pin (struct radeon_device *rdev);
drivers/gpu/drm/radeon/radeon_gart.c +18 -4
@@ -71,6 +71,7 @@int radeon_gart_table_ram_alloc (struct radeon_device *rdev) {void *ptr; +int ret; ptr =pci_alloc_consistent (rdev->pdev, rdev->gart.table_size, &rdev->gart.table_addr);
@@ -80,8 +81,16 @@ int radeon_gart_table_ram_alloc(struct radeon_device *rdev)#ifdef CONFIG_X86 if (rdev->family ==CHIP_RS400 || rdev->family ==CHIP_RS480 || rdev->family ==CHIP_RS690 || rdev->family ==CHIP_RS740 ) { -set_memory_uc ((unsigned long )ptr, + ret =set_memory_uc ((unsigned long )ptr, rdev->gart.table_size >>PAGE_SHIFT ); +if (ret) { +pci_free_consistent (rdev->pdev, rdev->gart.table_size, + (void *)rdev->gart.ptr, + rdev->gart.table_addr); + rdev->gart.ptr =NULL ; + rdev->gart.table_addr =0 ; +return ret; + } }#endif rdev->gart.ptr = ptr;
@@ -98,16 +107,20 @@ int radeon_gart_table_ram_alloc(struct radeon_device *rdev)* (r1xx-r3xx, non-pcie r4xx, rs400). These asics require the * gart table to be in system memory. */ -void radeon_gart_table_ram_free (struct radeon_device *rdev) +int radeon_gart_table_ram_free (struct radeon_device *rdev) { +int ret; +if (rdev->gart.ptr ==NULL ) { -return ; +return 0 ; }#ifdef CONFIG_X86 if (rdev->family ==CHIP_RS400 || rdev->family ==CHIP_RS480 || rdev->family ==CHIP_RS690 || rdev->family ==CHIP_RS740 ) { -set_memory_wb ((unsigned long )rdev->gart.ptr, + ret =set_memory_wb ((unsigned long )rdev->gart.ptr, rdev->gart.table_size >>PAGE_SHIFT ); +if (ret) +return ret; }#endif pci_free_consistent(rdev->pdev, rdev->gart.table_size,
@@ -115,6 +128,7 @@ void radeon_gart_table_ram_free(struct radeon_device *rdev) rdev->gart.table_addr); rdev->gart.ptr =NULL ; rdev->gart.table_addr =0 ; +return 0 ; }/**
Signature · 1 line
2.17.1
PATCH 2/2 drm/radeon: change call sites to handle return value properly.
In reply to Tianlin LiIdeally, the failure of set_memory_*() should be passed up the call stack, and callers should examine the failure and deal with it. Fix those call sites in drm/radeon to handle retval properly.
Since fini functions are always void, print errors for the failures.
--- drivers/gpu/drm/radeon/r100.c | 3 ++- drivers/gpu/drm/radeon/rs400.c | 3 ++- 2 files changed, 4 insertions(+), 2 deletions(-)
drivers/gpu/drm/radeon/r100.c +2 -1
@@ -706,7 +706,8 @@ void r100_pci_gart_fini(struct radeon_device *rdev) {radeon_gart_fini (rdev);r100_pci_gart_disable (rdev); -radeon_gart_table_ram_free (rdev); +if (radeon_gart_table_ram_free (rdev)) +DRM_ERROR ("radeon: failed free system ram for GART page table.\n" ); }int r100_irq_set (struct radeon_device *rdev)
drivers/gpu/drm/radeon/rs400.c +2 -1
@@ -210,7 +210,8 @@ void rs400_gart_fini(struct radeon_device *rdev) {radeon_gart_fini (rdev);rs400_gart_disable (rdev); -radeon_gart_table_ram_free (rdev); +if (radeon_gart_table_ram_free (rdev)) +DRM_ERROR ("radeon: failed free system ram for GART page table.\n" ); }#define RS400_PTE_UNSNOOPED (1 << 0)
Signature · 1 line
2.17.1
Christian König
maintainerAm 07.01.20 um 20:25 schrieb Tianlin Li:
Quoted 17 lines
Right now several architectures allow their set_memory_*() family of functions to fail, but callers may not be checking the return values. If set_memory_*() returns with an error, call-site assumptions may be infact wrong to assume that it would either succeed or not succeed at all. Ideally, the failure of set_memory_*() should be passed up the call stack, and callers should examine the failure and deal with it.
Need to fix the callers and add the
__must_checkattribute. They also may not provide any level of atomicity, in the sense that the memory protections may be left incomplete on failure. This issue likely has a few steps on effects architectures:
1)Have all callers of set_memory_*() helpers check the return value. 2)Add__must_checkto all set_memory_*() helpers so that new uses do not ignore the return value.
3)Add atomicity to the calls so that the memory protections aren't left in a partial state.
Quoted 2 lines
This series is part of step 1. Make drm/radeon check the return value of set_memory_*().
I'm a little hesitate merge that. This hardware is >15 years old and nobody of the developers have any system left to test this change on.
Would it be to much of a problem to just add something like: r = set_memory_*(); (void)r; /* Intentionally ignored */.
Apart from that certainly a good idea to add __must_check to the functions.
Regards,
Christian.
Quoted 11 lines
Tianlin Li (2):
drm/radeon: have the callers of set_memory_*() check the return value drm/radeon: change call sites to handle return value properly.drivers/gpu/drm/radeon/r100.c | 3 ++- drivers/gpu/drm/radeon/radeon.h | 2 +- drivers/gpu/drm/radeon/radeon_gart.c | 22 ++++++++++++++++++---- drivers/gpu/drm/radeon/rs400.c | 3 ++- 4 files changed, 23 insertions(+), 7 deletions(-)
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Quoted 21 lines · Christian König
Am 07.01.20 um 20:25 schrieb Tianlin Li:
Quoted 17 lines
Right now several architectures allow their set_memory_*() family of functions to fail, but callers may not be checking the return values. If set_memory_*() returns with an error, call-site assumptions may be infact wrong to assume that it would either succeed or not succeed at all. Ideally, the failure of set_memory_*() should be passed up the call stack, and callers should examine the failure and deal with it.
Need to fix the callers and add the
__must_checkattribute. They also may not provide any level of atomicity, in the sense that the memory protections may be left incomplete on failure. This issue likely has a few steps on effects architectures:
1)Have all callers of set_memory_*() helpers check the return value. 2)Add__must_checkto all set_memory_*() helpers so that new uses do not ignore the return value.
3)Add atomicity to the calls so that the memory protections aren't left in a partial state.
Quoted 6 lines
Quoted 2 lines
This series is part of step 1. Make drm/radeon check the return value of set_memory_*().
I'm a little hesitate merge that. This hardware is >15 years old and nobody of the developers have any system left to test this change on.
Would it be to much of a problem to just add something like: r = set_memory_*(); (void)r; /* Intentionally ignored */.
Thank you. I will fix that in patch 1 and remove patch 2 (since no need to fix the call sites to handle the retval).
Best regards,
Tianlin
Quoted 15 lines
Apart from that certainly a good idea to add
__must_checkto the functions.Regards,
Christian.Quoted 10 lines
Tianlin Li (2):
drm/radeon: have the callers of set_memory_*() check the return value drm/radeon: change call sites to handle return value properly.drivers/gpu/drm/radeon/r100.c | 3 ++- drivers/gpu/drm/radeon/radeon.h | 2 +- drivers/gpu/drm/radeon/radeon_gart.c | 22 ++++++++++++++++++---- drivers/gpu/drm/radeon/rs400.c | 3 ++- 4 files changed, 23 insertions(+), 7 deletions(-)
Quoted 18 lines · Christian König
Am 07.01.20 um 20:25 schrieb Tianlin Li:
Quoted 17 lines
Right now several architectures allow their set_memory_*() family of functions to fail, but callers may not be checking the return values. If set_memory_*() returns with an error, call-site assumptions may be infact wrong to assume that it would either succeed or not succeed at all. Ideally, the failure of set_memory_*() should be passed up the call stack, and callers should examine the failure and deal with it.
Need to fix the callers and add the
__must_checkattribute. They also may not provide any level of atomicity, in the sense that the memory protections may be left incomplete on failure. This issue likely has a few steps on effects architectures:
1)Have all callers of set_memory_*() helpers check the return value. 2)Add__must_checkto all set_memory_*() helpers so that new uses do not ignore the return value.
3)Add atomicity to the calls so that the memory protections aren't left in a partial state.
Quoted 5 lines
Quoted 2 lines
This series is part of step 1. Make drm/radeon check the return value of set_memory_*().
I'm a little hesitate merge that. This hardware is >15 years old and nobody of the developers have any system left to test this change on.
If that's true it should be removed from the tree. We need to be able to correctly make these kinds of changes in the kernel.
Quoted 2 lines
Would it be to much of a problem to just add something like: r = set_memory_*(); (void)r; /* Intentionally ignored */.
This seems like a bad idea -- we shouldn't be papering over failures like this when there is logic available to deal with it.
Quoted 1 line
Apart from that certainly a good idea to add
__must_checkto the functions.
Agreed!
-Kees
Signature · 1 line
Kees Cook
Quoted 20 lines · Kees Cook
Quoted 18 lines · Christian König
Am 07.01.20 um 20:25 schrieb Tianlin Li:
Quoted 17 lines
Right now several architectures allow their set_memory_*() family of functions to fail, but callers may not be checking the return values. If set_memory_*() returns with an error, call-site assumptions may be infact wrong to assume that it would either succeed or not succeed at all. Ideally, the failure of set_memory_*() should be passed up the call stack, and callers should examine the failure and deal with it.
Need to fix the callers and add the
__must_checkattribute. They also may not provide any level of atomicity, in the sense that the memory protections may be left incomplete on failure. This issue likely has a few steps on effects architectures:
1)Have all callers of set_memory_*() helpers check the return value. 2)Add__must_checkto all set_memory_*() helpers so that new uses do not ignore the return value.
3)Add atomicity to the calls so that the memory protections aren't left in a partial state.
Quoted 8 lines
Quoted 5 lines
Quoted 2 lines
This series is part of step 1. Make drm/radeon check the return value of set_memory_*().
I'm a little hesitate merge that. This hardware is >15 years old and nobody of the developers have any system left to test this change on.
If that's true it should be removed from the tree. We need to be able to correctly make these kinds of changes in the kernel.
This driver supports about 15 years of hardware generations. Newer cards are still prevalent, but the older stuff is less so. It still works and people use it based on feedback I've seen, but the older stuff has no active development at this point. This change just happens to target those older chips.
Alex
Quoted 11 lines
Quoted 3 lines
Would it be to much of a problem to just add something like: r = set_memory_*(); (void)r; /* Intentionally ignored */.
This seems like a bad idea -- we shouldn't be papering over failures like this when there is logic available to deal with it.
Quoted 1 line
Apart from that certainly a good idea to add
__must_checkto the functions.Agreed!
Quoted 8 lines
-Kees
--
Kees Cook
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Christian König
maintainerAm 08.01.20 um 18:51 schrieb Alex Deucher:
Quoted 24 lines · Kees Cook
Quoted 23 lines · Christian König
Quoted 22 lines
Am 07.01.20 um 20:25 schrieb Tianlin Li:
Quoted 17 lines
Right now several architectures allow their set_memory_*() family of functions to fail, but callers may not be checking the return values. If set_memory_*() returns with an error, call-site assumptions may be infact wrong to assume that it would either succeed or not succeed at all. Ideally, the failure of set_memory_*() should be passed up the call stack, and callers should examine the failure and deal with it.
Need to fix the callers and add the
__must_checkattribute. They also may not provide any level of atomicity, in the sense that the memory protections may be left incomplete on failure. This issue likely has a few steps on effects architectures:
1)Have all callers of set_memory_*() helpers check the return value. 2)Add__must_checkto all set_memory_*() helpers so that new uses do not ignore the return value.
3)Add atomicity to the calls so that the memory protections aren't left in a partial state.Quoted 2 lines
This series is part of step 1. Make drm/radeon check the return value of set_memory_*().
I'm a little hesitate merge that. This hardware is >15 years old and nobody of the developers have any system left to test this change on.
Quoted 7 lines
Quoted 2 lines
If that's true it should be removed from the tree. We need to be able to correctly make these kinds of changes in the kernel.
This driver supports about 15 years of hardware generations. Newer cards are still prevalent, but the older stuff is less so. It still works and people use it based on feedback I've seen, but the older stuff has no active development at this point. This change just happens to target those older chips.
Just a few weeks back we've got a mail from somebody using an integrated R128 in a laptop.
After a few mails back and force we figured out that his nearly 20 years old hardware was finally failing.
Up till that he was still successfully updating his kernel from time to time and the driver still worked. I find that pretty impressive.
Quoted 7 lines
Alex
Quoted 4 lines
Quoted 2 lines
Would it be to much of a problem to just add something like: r = set_memory_*(); (void)r; /* Intentionally ignored */.
This seems like a bad idea -- we shouldn't be papering over failures like this when there is logic available to deal with it.
Well I certainly agree to that, but we are talking about a call which happens only once during driver load/unload. If necessary we could also print an error when something goes wrong, but please no larger refactoring of return values and call paths.
It is perfectly possible that this call actually failed on somebodies hardware, but we never noticed because the driver still works fine. If we now handle the error it is possible that the module never loads and the user gets a black screen instead.
Regards,
Christian.
Quoted 4 lines
Quoted 4 lines
Quoted 2 lines
Apart from that certainly a good idea to add
__must_checkto the functions.Agreed!
Quoted 8 lines
Quoted 8 lines
-Kees
--
Kees Cook
_______________________________________________
dri-devel mailing list
[email protected]
https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.freedesktop.org%2Fmailman%2Flistinfo%2Fdri-devel&data=02%7C01%7Cchristian.koenig%40amd.com%7Ca542d384d54040b5b0b708d794636df1%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637141027080080147&sdata=EHFl6YOHmNp7gOqWsVmfoeD0jNirBTOGHcCP4efC%2FvE%3D&reserved=0
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
Hi
Am 09.01.20 um 11:15 schrieb Christian König:
Quoted 50 lines
Am 08.01.20 um 18:51 schrieb Alex Deucher:
Quoted 27 lines · Kees Cook
Quoted 26 lines · Christian König
Quoted 25 lines
Am 07.01.20 um 20:25 schrieb Tianlin Li:
Quoted 18 lines
Right now several architectures allow their set_memory_*() family of functions to fail, but callers may not be checking the return values. If set_memory_*() returns with an error, call-site assumptions may be infact wrong to assume that it would either succeed or not succeed at all. Ideally, the failure of set_memory_*() should be passed up the call stack, and callers should examine the failure and deal with it.
Need to fix the callers and add the
__must_checkattribute. They also may not provide any level of atomicity, in the sense that the memory protections may be left incomplete on failure. This issue likely has a few steps on effects architectures:
1)Have all callers of set_memory_*() helpers check the return value. 2)Add__must_checkto all set_memory_*() helpers so that new uses do not ignore the return value.
3)Add atomicity to the calls so that the memory protections aren't left
in a partial state.Quoted 3 lines
This series is part of step 1. Make drm/radeon check the return value of
set_memory_*().I'm a little hesitate merge that. This hardware is >15 years old and nobody
of the developers have any system left to test this change on.Quoted 7 lines
Quoted 2 lines
If that's true it should be removed from the tree. We need to be able to correctly make these kinds of changes in the kernel.
This driver supports about 15 years of hardware generations. Newer cards are still prevalent, but the older stuff is less so. It still works and people use it based on feedback I've seen, but the older stuff has no active development at this point. This change just happens to target those older chips.
Just a few weeks back we've got a mail from somebody using an integrated R128 in a laptop.
After a few mails back and force we figured out that his nearly 20 years old hardware was finally failing.
Up till that he was still successfully updating his kernel from time to time and the driver still worked. I find that pretty impressive.
Quoted 5 lines
Alex
Quoted 2 lines
Quoted 2 lines
Would it be to much of a problem to just add something like: r = set_memory_*(); (void)r; /* Intentionally ignored */.
Quoted 8 lines
Quoted 2 lines
Quoted 2 lines
This seems like a bad idea -- we shouldn't be papering over failures like this when there is logic available to deal with it.
Well I certainly agree to that, but we are talking about a call which happens only once during driver load/unload. If necessary we could also print an error when something goes wrong, but please no larger refactoring of return values and call paths.
IMHO radeon should be marked as orphaned or obsolete then.
Best regards
Thomas
Quoted 27 lines
It is perfectly possible that this call actually failed on somebodies hardware, but we never noticed because the driver still works fine. If we now handle the error it is possible that the module never loads and the user gets a black screen instead.
Regards,
Christian.Quoted 7 lines
Quoted 7 lines
Quoted 3 lines
Apart from that certainly a good idea to add
__must_checkto the functions.Agreed!
-Kees
Quoted 7 lines
Quoted 7 lines
Signature · 5 lines
Kees Cook _______________________________________________ dri-devel mailing list [email protected] https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.freedesktop.org%2Fmailman%2Flistinfo%2Fdri-devel&data=02%7C01%7Cchristian.koenig%40amd.com%7Ca542d384d54040b5b0b708d794636df1%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637141027080080147&sdata=EHFl6YOHmNp7gOqWsVmfoeD0jNirBTOGHcCP4efC%2FvE%3D&reserved=0_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
Signature · 6 lines
Thomas Zimmermann Graphics Driver Developer SUSE Software Solutions Germany GmbH Maxfeldstr. 5, 90409 Nürnberg, Germany (HRB 36809, AG Nürnberg) Geschäftsführer: Felix Imendörffer
Quoted 57 lines · Thomas Zimmermann
Hi
Am 09.01.20 um 11:15 schrieb Christian König:
Quoted 53 lines
Am 08.01.20 um 18:51 schrieb Alex Deucher:
Quoted 27 lines · Kees Cook
Quoted 26 lines · Christian König
Quoted 25 lines
Am 07.01.20 um 20:25 schrieb Tianlin Li:
Quoted 18 lines
Right now several architectures allow their set_memory_*() family of functions to fail, but callers may not be checking the return values. If set_memory_*() returns with an error, call-site assumptions may be infact wrong to assume that it would either succeed or not succeed at all. Ideally, the failure of set_memory_*() should be passed up the call stack, and callers should examine the failure and deal with it.
Need to fix the callers and add the
__must_checkattribute. They also may not provide any level of atomicity, in the sense that the memory protections may be left incomplete on failure. This issue likely has a few steps on effects architectures:
1)Have all callers of set_memory_*() helpers check the return value. 2)Add__must_checkto all set_memory_*() helpers so that new uses do not ignore the return value.
3)Add atomicity to the calls so that the memory protections aren't left
in a partial state.Quoted 3 lines
This series is part of step 1. Make drm/radeon check the return value of
set_memory_*().I'm a little hesitate merge that. This hardware is >15 years old and nobody
of the developers have any system left to test this change on.Quoted 7 lines
Quoted 2 lines
If that's true it should be removed from the tree. We need to be able to correctly make these kinds of changes in the kernel.
This driver supports about 15 years of hardware generations. Newer cards are still prevalent, but the older stuff is less so. It still works and people use it based on feedback I've seen, but the older stuff has no active development at this point. This change just happens to target those older chips.
Just a few weeks back we've got a mail from somebody using an integrated R128 in a laptop.
After a few mails back and force we figured out that his nearly 20 years old hardware was finally failing.
Up till that he was still successfully updating his kernel from time to time and the driver still worked. I find that pretty impressive.
Quoted 7 lines
Alex
Quoted 4 lines
Quoted 2 lines
Would it be to much of a problem to just add something like: r = set_memory_*(); (void)r; /* Intentionally ignored */.
This seems like a bad idea -- we shouldn't be papering over failure like this when there is logic available to deal with it.
Quoted 7 lines
Quoted 5 lines
Well I certainly agree to that, but we are talking about a call which happens only once during driver load/unload. If necessary we could also print an error when something goes wrong, but please no larger refactoring of return values and call paths.
IMHO radeon should be marked as orphaned or obsolete then.
As I said this covers about 15-17 years of GPUs (~60 asic families). The older stuff is hard to test these days because it's PCI or AGP hardware. So far it works for most people. The newer stuff is still tested as used regularly.
Alex
Quoted 32 lines
Best regards
ThomasQuoted 23 lines
It is perfectly possible that this call actually failed on somebodies hardware, but we never noticed because the driver still works fine. If we now handle the error it is possible that the module never loads and the user gets a black screen instead.
Regards,
Christian.Quoted 7 lines
Quoted 7 lines
Quoted 3 lines
Apart from that certainly a good idea to add
__must_checkto the functions.Agreed!
-Kees
Quoted 7 lines
Quoted 7 lines
--
Kees Cook
_______________________________________________
dri-devel mailing list
[email protected]
https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.freedesktop.org%2Fmailman%2Flistinfo%2Fdri-devel&data=02%7C01%7Cchristian.koenig%40amd.com%7Ca542d384d54040b5b0b708d794636df1%7C3dd8961fe4884e608e11a82d994e183d%7C0%7C0%7C637141027080080147&sdata=EHFl6YOHmNp7gOqWsVmfoeD0jNirBTOGHcCP4efC%2FvE%3D&reserved=0Quoted 4 lines
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
Quoted 8 lines
--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
(HRB 36809, AG Nürnberg)
Geschäftsführer: Felix Imendörffer
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel