tldr; DMA buffers aren't normal memory, expecting that you can use them like that (like calling get_user_pages works, or that they're accounting like any other normal memory) cannot be guaranteed.
Since some userspace only runs on integrated devices, where all buffers are actually all resident system memory, there's a huge temptation to assume that a struct page is always present and useable like for any more pagecache backed mmap. This has the potential to result in a uapi nightmare.
To stop this gap require that DMA buffer mmaps are VM_SPECIAL, which blocks get_user_pages and all the other struct page based infrastructure for everyone. In spirit this is the uapi counterpart to the kernel-internal CONFIG_DMABUF_DEBUG.
Motivated by a recent patch which wanted to swich the system dma-buf heap to vm_insert_page instead of vm_insert_pfn.
References: https://lore.kernel.org/lkml/CAKMK7uHi+mG0z0HUmNt13QCCvutuRVjpcR0NjRL12k-WbW... Cc: Jason Gunthorpe jgg@ziepe.ca Cc: Suren Baghdasaryan surenb@google.com Cc: Matthew Wilcox willy@infradead.org Cc: John Stultz john.stultz@linaro.org Signed-off-by: Daniel Vetter daniel.vetter@intel.com Cc: Sumit Semwal sumit.semwal@linaro.org Cc: "Christian König" christian.koenig@amd.com Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org --- drivers/dma-buf/dma-buf.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-)
diff --git a/drivers/dma-buf/dma-buf.c b/drivers/dma-buf/dma-buf.c index f264b70c383e..d3081fc07056 100644 --- a/drivers/dma-buf/dma-buf.c +++ b/drivers/dma-buf/dma-buf.c @@ -127,6 +127,7 @@ static struct file_system_type dma_buf_fs_type = { static int dma_buf_mmap_internal(struct file *file, struct vm_area_struct *vma) { struct dma_buf *dmabuf; + int ret;
if (!is_dma_buf_file(file)) return -EINVAL; @@ -142,7 +143,11 @@ static int dma_buf_mmap_internal(struct file *file, struct vm_area_struct *vma) dmabuf->size >> PAGE_SHIFT) return -EINVAL;
- return dmabuf->ops->mmap(dmabuf, vma); + ret = dmabuf->ops->mmap(dmabuf, vma); + + WARN_ON(!(vma->vm_flags & VM_SPECIAL)); + + return ret; }
static loff_t dma_buf_llseek(struct file *file, loff_t offset, int whence) @@ -1244,6 +1249,8 @@ EXPORT_SYMBOL_GPL(dma_buf_end_cpu_access); int dma_buf_mmap(struct dma_buf *dmabuf, struct vm_area_struct *vma, unsigned long pgoff) { + int ret; + if (WARN_ON(!dmabuf || !vma)) return -EINVAL;
@@ -1264,7 +1271,11 @@ int dma_buf_mmap(struct dma_buf *dmabuf, struct vm_area_struct *vma, vma_set_file(vma, dmabuf->file); vma->vm_pgoff = pgoff;
- return dmabuf->ops->mmap(dmabuf, vma); + ret = dmabuf->ops->mmap(dmabuf, vma); + + WARN_ON(!(vma->vm_flags & VM_SPECIAL)); + + return ret; } EXPORT_SYMBOL_GPL(dma_buf_mmap);
On Wed, Feb 03, 2021 at 10:19:48PM +0100, Daniel Vetter wrote:
tldr; DMA buffers aren't normal memory, expecting that you can use them like that (like calling get_user_pages works, or that they're accounting like any other normal memory) cannot be guaranteed.
Since some userspace only runs on integrated devices, where all buffers are actually all resident system memory, there's a huge temptation to assume that a struct page is always present and useable like for any more pagecache backed mmap. This has the potential to result in a uapi nightmare.
To stop this gap require that DMA buffer mmaps are VM_SPECIAL, which blocks get_user_pages and all the other struct page based infrastructure for everyone. In spirit this is the uapi counterpart to the kernel-internal CONFIG_DMABUF_DEBUG.
Fast gup needs the special flag set on the PTE as well.. Feels weird to have a special VMA without also having special PTEs?
Jason
On Thu, Feb 4, 2021 at 5:13 PM Jason Gunthorpe jgg@ziepe.ca wrote:
On Wed, Feb 03, 2021 at 10:19:48PM +0100, Daniel Vetter wrote:
tldr; DMA buffers aren't normal memory, expecting that you can use them like that (like calling get_user_pages works, or that they're accounting like any other normal memory) cannot be guaranteed.
Since some userspace only runs on integrated devices, where all buffers are actually all resident system memory, there's a huge temptation to assume that a struct page is always present and useable like for any more pagecache backed mmap. This has the potential to result in a uapi nightmare.
To stop this gap require that DMA buffer mmaps are VM_SPECIAL, which blocks get_user_pages and all the other struct page based infrastructure for everyone. In spirit this is the uapi counterpart to the kernel-internal CONFIG_DMABUF_DEBUG.
Fast gup needs the special flag set on the PTE as well.. Feels weird to have a special VMA without also having special PTEs?
There's kinda no convenient & cheap way to check for the pte_special flag. This here should at least catch accidental misuse, people building their own ptes we can't stop. Maybe we should exclude VM_MIXEDMAP to catch vm_insert_page in one of these.
Hm looking at code I think we need to require VM_PFNMAP here to stop vm_insert_page. And looking at the various functions, that seems to be required (and I guess VM_IO is more for really funky architectures where io-space is somewhere else?). I guess I should check for VM_PFNMAP instead of VM_SPECIAL? -Daniel
On Thu, Feb 04, 2021 at 06:16:27PM +0100, Daniel Vetter wrote:
On Thu, Feb 4, 2021 at 5:13 PM Jason Gunthorpe jgg@ziepe.ca wrote:
On Wed, Feb 03, 2021 at 10:19:48PM +0100, Daniel Vetter wrote:
tldr; DMA buffers aren't normal memory, expecting that you can use them like that (like calling get_user_pages works, or that they're accounting like any other normal memory) cannot be guaranteed.
Since some userspace only runs on integrated devices, where all buffers are actually all resident system memory, there's a huge temptation to assume that a struct page is always present and useable like for any more pagecache backed mmap. This has the potential to result in a uapi nightmare.
To stop this gap require that DMA buffer mmaps are VM_SPECIAL, which blocks get_user_pages and all the other struct page based infrastructure for everyone. In spirit this is the uapi counterpart to the kernel-internal CONFIG_DMABUF_DEBUG.
Fast gup needs the special flag set on the PTE as well.. Feels weird to have a special VMA without also having special PTEs?
There's kinda no convenient & cheap way to check for the pte_special flag. This here should at least catch accidental misuse, people building their own ptes we can't stop. Maybe we should exclude VM_MIXEDMAP to catch vm_insert_page in one of these.
Hm looking at code I think we need to require VM_PFNMAP here to stop vm_insert_page. And looking at the various functions, that seems to be required (and I guess VM_IO is more for really funky architectures where io-space is somewhere else?). I guess I should check for VM_PFNMAP instead of VM_SPECIAL?
Well, you said the goal was to block GUP usage, that won't happen without the PTE special flag, at least on x86
So, really, what you are saying is all dmabuf users should always use vmf_insert_pfn_prot() or something similar - and never insert_page/etc?
It might make sense to check the vma flags in all the insert paths, eg vm_insert_page() can't work with VMAs that should not have struct pages in them (eg VM_SPECIAl, VM_PFNMAP, !VM_MIXEMAP if I understand it right)
At least as some VM debug option
Jason
On Thu, Feb 4, 2021 at 7:38 PM Jason Gunthorpe jgg@ziepe.ca wrote:
On Thu, Feb 04, 2021 at 06:16:27PM +0100, Daniel Vetter wrote:
On Thu, Feb 4, 2021 at 5:13 PM Jason Gunthorpe jgg@ziepe.ca wrote:
On Wed, Feb 03, 2021 at 10:19:48PM +0100, Daniel Vetter wrote:
tldr; DMA buffers aren't normal memory, expecting that you can use them like that (like calling get_user_pages works, or that they're accounting like any other normal memory) cannot be guaranteed.
Since some userspace only runs on integrated devices, where all buffers are actually all resident system memory, there's a huge temptation to assume that a struct page is always present and useable like for any more pagecache backed mmap. This has the potential to result in a uapi nightmare.
To stop this gap require that DMA buffer mmaps are VM_SPECIAL, which blocks get_user_pages and all the other struct page based infrastructure for everyone. In spirit this is the uapi counterpart to the kernel-internal CONFIG_DMABUF_DEBUG.
Fast gup needs the special flag set on the PTE as well.. Feels weird to have a special VMA without also having special PTEs?
There's kinda no convenient & cheap way to check for the pte_special flag. This here should at least catch accidental misuse, people building their own ptes we can't stop. Maybe we should exclude VM_MIXEDMAP to catch vm_insert_page in one of these.
Hm looking at code I think we need to require VM_PFNMAP here to stop vm_insert_page. And looking at the various functions, that seems to be required (and I guess VM_IO is more for really funky architectures where io-space is somewhere else?). I guess I should check for VM_PFNMAP instead of VM_SPECIAL?
Well, you said the goal was to block GUP usage, that won't happen without the PTE special flag, at least on x86
So, really, what you are saying is all dmabuf users should always use vmf_insert_pfn_prot() or something similar - and never insert_page/etc?
It might make sense to check the vma flags in all the insert paths, eg vm_insert_page() can't work with VMAs that should not have struct pages in them (eg VM_SPECIAl, VM_PFNMAP, !VM_MIXEMAP if I understand it right)
Well that's what I've done, and it /looks/ like all the checks are there already, as long as we use VM_PFNMAP. vm_insert_page tries to auto-add VM_MIXEDMAP, but bails out with a BUG_ON if VM_PFNMAP is set. And all the vm_insert_pfn_prot/remap_pfn_range functions require (or set) VM_PFNMAP.
So I think just checking for VM_PFNMAP after the vma is set up should be enough to guarantee we'll only have pte_special ptes in there, ever. But I'm not sure, this stuff all isn't really documented much and the code is sometimes a maze (to me at least).
At least as some VM debug option
Seems to be there already unconditionally. -Daniel
On Thu, Feb 04, 2021 at 08:59:59PM +0100, Daniel Vetter wrote:
So I think just checking for VM_PFNMAP after the vma is set up should be enough to guarantee we'll only have pte_special ptes in there, ever. But I'm not sure, this stuff all isn't really documented much and the code is sometimes a maze (to me at least).
Yes, that makes sense. VM_PFNMAP and !VM_MIXEDMAP seems like the right check after the VMA is populated
But how do you stuff special pfns into a VMA outside the fault handler?
Jason
On Thu, Feb 4, 2021 at 9:09 PM Jason Gunthorpe jgg@ziepe.ca wrote:
On Thu, Feb 04, 2021 at 08:59:59PM +0100, Daniel Vetter wrote:
So I think just checking for VM_PFNMAP after the vma is set up should be enough to guarantee we'll only have pte_special ptes in there, ever. But I'm not sure, this stuff all isn't really documented much and the code is sometimes a maze (to me at least).
Yes, that makes sense. VM_PFNMAP and !VM_MIXEDMAP seems like the right check after the VMA is populated
But how do you stuff special pfns into a VMA outside the fault handler?
Many drivers we have don't have dynamic buffer management (kinda overkill for a few framebuffers on a display-only IP block), so the just remap_pfn_range on ->mmap, and don't have a fault handler at all.
Or I'm not understanding what you're asking? -Daniel
On Thu, Feb 04, 2021 at 09:19:57PM +0100, Daniel Vetter wrote:
On Thu, Feb 4, 2021 at 9:09 PM Jason Gunthorpe jgg@ziepe.ca wrote:
On Thu, Feb 04, 2021 at 08:59:59PM +0100, Daniel Vetter wrote:
So I think just checking for VM_PFNMAP after the vma is set up should be enough to guarantee we'll only have pte_special ptes in there, ever. But I'm not sure, this stuff all isn't really documented much and the code is sometimes a maze (to me at least).
Yes, that makes sense. VM_PFNMAP and !VM_MIXEDMAP seems like the right check after the VMA is populated
But how do you stuff special pfns into a VMA outside the fault handler?
Many drivers we have don't have dynamic buffer management (kinda overkill for a few framebuffers on a display-only IP block), so the just remap_pfn_range on ->mmap, and don't have a fault handler at all.
remap_pfn_range() makes sense, do you expect drivers using struct page backed memory to call that as well?
Jason
On Thu, Feb 4, 2021 at 9:59 PM Jason Gunthorpe jgg@ziepe.ca wrote:
On Thu, Feb 04, 2021 at 09:19:57PM +0100, Daniel Vetter wrote:
On Thu, Feb 4, 2021 at 9:09 PM Jason Gunthorpe jgg@ziepe.ca wrote:
On Thu, Feb 04, 2021 at 08:59:59PM +0100, Daniel Vetter wrote:
So I think just checking for VM_PFNMAP after the vma is set up should be enough to guarantee we'll only have pte_special ptes in there, ever. But I'm not sure, this stuff all isn't really documented much and the code is sometimes a maze (to me at least).
Yes, that makes sense. VM_PFNMAP and !VM_MIXEDMAP seems like the right check after the VMA is populated
But how do you stuff special pfns into a VMA outside the fault handler?
Many drivers we have don't have dynamic buffer management (kinda overkill for a few framebuffers on a display-only IP block), so the just remap_pfn_range on ->mmap, and don't have a fault handler at all.
remap_pfn_range() makes sense, do you expect drivers using struct page backed memory to call that as well?
All the ones using CMA through dma_alloc_coherent end up in there with the dma_mmap_wc function. So yeah we have tons already.
The drivers with dynamic memory management all use vm_insert_pfn, even when the buffer is in system memory and struct page backed. I think those are the two cases. There's another mmap in drm/i915, but that should never leave intel-specific userspace, and I think we're also phasing it out somewhat. Either way, should never show up in a shared buffer usecase, ever, so I think we can ignore it. -Daniel
Am 04.02.21 um 19:38 schrieb Jason Gunthorpe:
On Thu, Feb 04, 2021 at 06:16:27PM +0100, Daniel Vetter wrote:
On Thu, Feb 4, 2021 at 5:13 PM Jason Gunthorpe jgg@ziepe.ca wrote:
On Wed, Feb 03, 2021 at 10:19:48PM +0100, Daniel Vetter wrote:
tldr; DMA buffers aren't normal memory, expecting that you can use them like that (like calling get_user_pages works, or that they're accounting like any other normal memory) cannot be guaranteed.
Since some userspace only runs on integrated devices, where all buffers are actually all resident system memory, there's a huge temptation to assume that a struct page is always present and useable like for any more pagecache backed mmap. This has the potential to result in a uapi nightmare.
To stop this gap require that DMA buffer mmaps are VM_SPECIAL, which blocks get_user_pages and all the other struct page based infrastructure for everyone. In spirit this is the uapi counterpart to the kernel-internal CONFIG_DMABUF_DEBUG.
Fast gup needs the special flag set on the PTE as well.. Feels weird to have a special VMA without also having special PTEs?
There's kinda no convenient & cheap way to check for the pte_special flag. This here should at least catch accidental misuse, people building their own ptes we can't stop. Maybe we should exclude VM_MIXEDMAP to catch vm_insert_page in one of these.
Hm looking at code I think we need to require VM_PFNMAP here to stop vm_insert_page. And looking at the various functions, that seems to be required (and I guess VM_IO is more for really funky architectures where io-space is somewhere else?). I guess I should check for VM_PFNMAP instead of VM_SPECIAL?
Well, you said the goal was to block GUP usage, that won't happen without the PTE special flag, at least on x86
When is that special flag being set?
So, really, what you are saying is all dmabuf users should always use vmf_insert_pfn_prot() or something similar - and never insert_page/etc?
Exactly, yes.
Christian.
It might make sense to check the vma flags in all the insert paths, eg vm_insert_page() can't work with VMAs that should not have struct pages in them (eg VM_SPECIAl, VM_PFNMAP, !VM_MIXEMAP if I understand it right)
At least as some VM debug option
Jason
tldr; DMA buffers aren't normal memory, expecting that you can use them like that (like calling get_user_pages works, or that they're accounting like any other normal memory) cannot be guaranteed.
Since some userspace only runs on integrated devices, where all buffers are actually all resident system memory, there's a huge temptation to assume that a struct page is always present and useable like for any more pagecache backed mmap. This has the potential to result in a uapi nightmare.
To stop this gap require that DMA buffer mmaps are VM_PFNMAP, which blocks get_user_pages and all the other struct page based infrastructure for everyone. In spirit this is the uapi counterpart to the kernel-internal CONFIG_DMABUF_DEBUG.
Motivated by a recent patch which wanted to swich the system dma-buf heap to vm_insert_page instead of vm_insert_pfn.
v2:
Jason brought up that we also want to guarantee that all ptes have the pte_special flag set, to catch fast get_user_pages (on architectures that support this). Allowing VM_MIXEDMAP (like VM_SPECIAL does) would still allow vm_insert_page, but limiting to VM_PFNMAP will catch that.
From auditing the various functions to insert pfn pte entires
(vm_insert_pfn_prot, remap_pfn_range and all it's callers like dma_mmap_wc) it looks like VM_PFNMAP is already required anyway, so this should be the correct flag to check for.
References: https://lore.kernel.org/lkml/CAKMK7uHi+mG0z0HUmNt13QCCvutuRVjpcR0NjRL12k-WbW... Cc: Jason Gunthorpe jgg@ziepe.ca Cc: Suren Baghdasaryan surenb@google.com Cc: Matthew Wilcox willy@infradead.org Cc: John Stultz john.stultz@linaro.org Signed-off-by: Daniel Vetter daniel.vetter@intel.com Cc: Sumit Semwal sumit.semwal@linaro.org Cc: "Christian König" christian.koenig@amd.com Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org --- drivers/dma-buf/dma-buf.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-)
diff --git a/drivers/dma-buf/dma-buf.c b/drivers/dma-buf/dma-buf.c index f264b70c383e..06cb1d2e9fdc 100644 --- a/drivers/dma-buf/dma-buf.c +++ b/drivers/dma-buf/dma-buf.c @@ -127,6 +127,7 @@ static struct file_system_type dma_buf_fs_type = { static int dma_buf_mmap_internal(struct file *file, struct vm_area_struct *vma) { struct dma_buf *dmabuf; + int ret;
if (!is_dma_buf_file(file)) return -EINVAL; @@ -142,7 +143,11 @@ static int dma_buf_mmap_internal(struct file *file, struct vm_area_struct *vma) dmabuf->size >> PAGE_SHIFT) return -EINVAL;
- return dmabuf->ops->mmap(dmabuf, vma); + ret = dmabuf->ops->mmap(dmabuf, vma); + + WARN_ON(!(vma->vm_flags & VM_PFNMAP)); + + return ret; }
static loff_t dma_buf_llseek(struct file *file, loff_t offset, int whence) @@ -1244,6 +1249,8 @@ EXPORT_SYMBOL_GPL(dma_buf_end_cpu_access); int dma_buf_mmap(struct dma_buf *dmabuf, struct vm_area_struct *vma, unsigned long pgoff) { + int ret; + if (WARN_ON(!dmabuf || !vma)) return -EINVAL;
@@ -1264,7 +1271,11 @@ int dma_buf_mmap(struct dma_buf *dmabuf, struct vm_area_struct *vma, vma_set_file(vma, dmabuf->file); vma->vm_pgoff = pgoff;
- return dmabuf->ops->mmap(dmabuf, vma); + ret = dmabuf->ops->mmap(dmabuf, vma); + + WARN_ON(!(vma->vm_flags & VM_PFNMAP)); + + return ret; } EXPORT_SYMBOL_GPL(dma_buf_mmap);
Am 05.02.21 um 14:41 schrieb Daniel Vetter:
tldr; DMA buffers aren't normal memory, expecting that you can use them like that (like calling get_user_pages works, or that they're accounting like any other normal memory) cannot be guaranteed.
Since some userspace only runs on integrated devices, where all buffers are actually all resident system memory, there's a huge temptation to assume that a struct page is always present and useable like for any more pagecache backed mmap. This has the potential to result in a uapi nightmare.
To stop this gap require that DMA buffer mmaps are VM_PFNMAP, which blocks get_user_pages and all the other struct page based infrastructure for everyone. In spirit this is the uapi counterpart to the kernel-internal CONFIG_DMABUF_DEBUG.
Motivated by a recent patch which wanted to swich the system dma-buf heap to vm_insert_page instead of vm_insert_pfn.
v2:
Jason brought up that we also want to guarantee that all ptes have the pte_special flag set, to catch fast get_user_pages (on architectures that support this). Allowing VM_MIXEDMAP (like VM_SPECIAL does) would still allow vm_insert_page, but limiting to VM_PFNMAP will catch that.
From auditing the various functions to insert pfn pte entires (vm_insert_pfn_prot, remap_pfn_range and all it's callers like dma_mmap_wc) it looks like VM_PFNMAP is already required anyway, so this should be the correct flag to check for.
References: https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flore.kerne... Cc: Jason Gunthorpe jgg@ziepe.ca Cc: Suren Baghdasaryan surenb@google.com Cc: Matthew Wilcox willy@infradead.org Cc: John Stultz john.stultz@linaro.org Signed-off-by: Daniel Vetter daniel.vetter@intel.com Cc: Sumit Semwal sumit.semwal@linaro.org Cc: "Christian König" christian.koenig@amd.com Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org
Acked-by: Christian König christian.koenig@amd.com
drivers/dma-buf/dma-buf.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-)
diff --git a/drivers/dma-buf/dma-buf.c b/drivers/dma-buf/dma-buf.c index f264b70c383e..06cb1d2e9fdc 100644 --- a/drivers/dma-buf/dma-buf.c +++ b/drivers/dma-buf/dma-buf.c @@ -127,6 +127,7 @@ static struct file_system_type dma_buf_fs_type = { static int dma_buf_mmap_internal(struct file *file, struct vm_area_struct *vma) { struct dma_buf *dmabuf;
int ret;
if (!is_dma_buf_file(file)) return -EINVAL;
@@ -142,7 +143,11 @@ static int dma_buf_mmap_internal(struct file *file, struct vm_area_struct *vma) dmabuf->size >> PAGE_SHIFT) return -EINVAL;
- return dmabuf->ops->mmap(dmabuf, vma);
ret = dmabuf->ops->mmap(dmabuf, vma);
WARN_ON(!(vma->vm_flags & VM_PFNMAP));
return ret; }
static loff_t dma_buf_llseek(struct file *file, loff_t offset, int whence)
@@ -1244,6 +1249,8 @@ EXPORT_SYMBOL_GPL(dma_buf_end_cpu_access); int dma_buf_mmap(struct dma_buf *dmabuf, struct vm_area_struct *vma, unsigned long pgoff) {
- int ret;
- if (WARN_ON(!dmabuf || !vma)) return -EINVAL;
@@ -1264,7 +1271,11 @@ int dma_buf_mmap(struct dma_buf *dmabuf, struct vm_area_struct *vma, vma_set_file(vma, dmabuf->file); vma->vm_pgoff = pgoff;
- return dmabuf->ops->mmap(dmabuf, vma);
- ret = dmabuf->ops->mmap(dmabuf, vma);
- WARN_ON(!(vma->vm_flags & VM_PFNMAP));
- return ret; } EXPORT_SYMBOL_GPL(dma_buf_mmap);
Greeting,
FYI, we noticed the following commit (built with gcc-9):
commit: b922393a2c5c06b9262c46d5f7d3682a5b9f70d4 ("RFC: dma-buf: Require VM_SPECIAL vma for mmap") https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git Daniel-Vetter/RFC-dma-buf-Require-VM_SPECIAL-vma-for-mmap/20210204-052229
in testcase: igt version: igt-x86_64-e230cd8d-1_20210106 with following parameters:
group: group-06 ucode: 0xe2
on test machine: 8 threads Intel(R) Core(TM) i7-6700 CPU @ 3.40GHz with 28G memory
caused below changes (please refer to attached dmesg/kmsg for entire log/backtrace):
If you fix the issue, kindly add following tag Reported-by: kernel test robot oliver.sang@intel.com
[ 615.523538] WARNING: CPU: 1 PID: 6792 at drivers/dma-buf/dma-buf.c:136 dma_buf_mmap_internal (kbuild/src/consumer/drivers/dma-buf/dma-buf.c:136 (discriminator 1)) [ 615.528827] Subtest atomic-pipe-F-tiling-yf: SKIP [ 615.529645] Modules linked in: i915 drm_kms_helper drm ipmi_devintf ipmi_msghandler btrfs [ 615.538950] [ 615.543650] blake2b_generic xor zstd_compress raid6_pq libcrc32c sd_mod t10_pi [ 615.553212] 2021-02-04 20:38:34 build/tests/prime_mmap_kms --run-subtest buffer-sharing [ 615.553344] sg [ 615.560620] [ 615.568616] intel_rapl_msr intel_rapl_common x86_pkg_temp_thermal intel_powerclamp coretemp kvm_intel intel_gtt kvm syscopyarea irqbypass crct10dif_pclmul sysfillrect crc32_pclmul crc32c_intel sysimgblt ahci ghash_clmulni_intel mei_wdt dell_wmi libahci fb_sys_fops rapl mei_me dell_smbios intel_cstate wmi_bmof sparse_keymap dell_wmi_descriptor i2c_i801 dcdbas i2c_smbus libata intel_uncore mei joydev intel_pch_thermal wmi video intel_pmc_core acpi_pad ip_tables [last unloaded: drm] [ 615.614440] CPU: 1 PID: 6792 Comm: prime_mmap_kms Tainted: G I 5.11.0-rc1-00007-gb922393a2c5c #1 [ 615.624521] Hardware name: Dell Inc. OptiPlex 7040/0Y7WYT, BIOS 1.2.8 01/26/2016 [ 615.631923] RIP: 0010:dma_buf_mmap_internal (kbuild/src/consumer/drivers/dma-buf/dma-buf.c:136 (discriminator 1)) [ 615.636986] Code: 46 08 48 2b 06 48 c1 e8 0c 48 c1 ea 0c 48 03 86 98 00 00 00 48 39 d0 77 1b e8 d6 03 69 00 48 f7 43 50 00 44 04 10 74 02 5b c3 <0f> 0b 5b c3 b8 ea ff ff ff c3 b8 ea ff ff ff 5b c3 66 66 2e 0f 1f All code ======== 0: 46 08 48 2b rex.RX or %r9b,0x2b(%rax) 4: 06 (bad) 5: 48 c1 e8 0c shr $0xc,%rax 9: 48 c1 ea 0c shr $0xc,%rdx d: 48 03 86 98 00 00 00 add 0x98(%rsi),%rax 14: 48 39 d0 cmp %rdx,%rax 17: 77 1b ja 0x34 19: e8 d6 03 69 00 callq 0x6903f4 1e: 48 f7 43 50 00 44 04 testq $0x10044400,0x50(%rbx) 25: 10 26: 74 02 je 0x2a 28: 5b pop %rbx 29: c3 retq 2a:* 0f 0b ud2 <-- trapping instruction 2c: 5b pop %rbx 2d: c3 retq 2e: b8 ea ff ff ff mov $0xffffffea,%eax 33: c3 retq 34: b8 ea ff ff ff mov $0xffffffea,%eax 39: 5b pop %rbx 3a: c3 retq 3b: 66 data16 3c: 66 data16 3d: 2e cs 3e: 0f .byte 0xf 3f: 1f (bad)
Code starting with the faulting instruction =========================================== 0: 0f 0b ud2 2: 5b pop %rbx 3: c3 retq 4: b8 ea ff ff ff mov $0xffffffea,%eax 9: c3 retq a: b8 ea ff ff ff mov $0xffffffea,%eax f: 5b pop %rbx 10: c3 retq 11: 66 data16 12: 66 data16 13: 2e cs 14: 0f .byte 0xf 15: 1f (bad) [ 615.655745] RSP: 0018:ffffc900013ffd98 EFLAGS: 00010246 [ 615.660968] RAX: 0000000000000000 RBX: ffff8887580ee3e8 RCX: 00000000601c5b4b [ 615.668101] RDX: 00007f840ba00000 RSI: 0000000000000001 RDI: ffff888757b34800 [ 615.675232] RBP: 00000000000000d1 R08: ffff888757b34800 R09: 0000000000015428 [ 615.682362] R10: 00007f840bbfe000 R11: ffff8882013c28c0 R12: ffff8882013c28b8 [ 615.689493] R13: ffff88874588eec0 R14: ffff8887580ee3e8 R15: ffff8882013c2898 [ 615.696624] FS: 00007f840816cc00(0000) GS:ffff888759c40000(0000) knlGS:0000000000000000 [ 615.704710] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 615.710460] CR2: 00007f840bc742a0 CR3: 00000001c9ee0001 CR4: 00000000003706e0 [ 615.717601] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 615.724744] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 [ 615.731893] Call Trace: [ 615.734356] mmap_region (kbuild/src/consumer/mm/mmap.c:1808) [ 615.737935] do_mmap (kbuild/src/consumer/mm/mmap.c:1583) [ 615.741161] ? security_mmap_file (kbuild/src/consumer/security/security.c:1523) [ 615.745343] vm_mmap_pgoff (kbuild/src/consumer/include/linux/mmap_lock.h:104 kbuild/src/consumer/mm/util.c:521) [ 615.749005] ksys_mmap_pgoff (kbuild/src/consumer/mm/mmap.c:1634) [ 615.752940] do_syscall_64 (kbuild/src/consumer/arch/x86/entry/common.c:46) [ 615.756515] entry_SYSCALL_64_after_hwframe (kbuild/src/consumer/arch/x86/entry/entry_64.S:127) [ 615.761565] RIP: 0033:0x7f840bb09133 [ 615.765138] Code: 54 41 89 d4 55 48 89 fd 53 4c 89 cb 48 85 ff 74 56 49 89 d9 45 89 f8 45 89 f2 44 89 e2 4c 89 ee 48 89 ef b8 09 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 7d 5b 5d 41 5c 41 5d 41 5e 41 5f c3 66 2e 0f All code ======== 0: 54 push %rsp 1: 41 89 d4 mov %edx,%r12d 4: 55 push %rbp 5: 48 89 fd mov %rdi,%rbp 8: 53 push %rbx 9: 4c 89 cb mov %r9,%rbx c: 48 85 ff test %rdi,%rdi f: 74 56 je 0x67 11: 49 89 d9 mov %rbx,%r9 14: 45 89 f8 mov %r15d,%r8d 17: 45 89 f2 mov %r14d,%r10d 1a: 44 89 e2 mov %r12d,%edx 1d: 4c 89 ee mov %r13,%rsi 20: 48 89 ef mov %rbp,%rdi 23: b8 09 00 00 00 mov $0x9,%eax 28: 0f 05 syscall 2a:* 48 3d 00 f0 ff ff cmp $0xfffffffffffff000,%rax <-- trapping instruction 30: 77 7d ja 0xaf 32: 5b pop %rbx 33: 5d pop %rbp 34: 41 5c pop %r12 36: 41 5d pop %r13 38: 41 5e pop %r14 3a: 41 5f pop %r15 3c: c3 retq 3d: 66 data16 3e: 2e cs 3f: 0f .byte 0xf
Code starting with the faulting instruction =========================================== 0: 48 3d 00 f0 ff ff cmp $0xfffffffffffff000,%rax 6: 77 7d ja 0x85 8: 5b pop %rbx 9: 5d pop %rbp a: 41 5c pop %r12 c: 41 5d pop %r13 e: 41 5e pop %r14 10: 41 5f pop %r15 12: c3 retq 13: 66 data16 14: 2e cs 15: 0f .byte 0xf
To reproduce:
git clone https://github.com/intel/lkp-tests.git cd lkp-tests bin/lkp install job.yaml # job file is attached in this email bin/lkp split-job --compatible job.yaml bin/lkp run compatible-job.yaml
Thanks, Oliver Sang
dri-devel@lists.freedesktop.org
-
Christian König -
Daniel Vetter -
Jason Gunthorpe -
kernel test robot