Re: [PATCH 1/2] drm/vgem: Do not allocate backing shmemfs file for an import dmabuf object
Am 07.07.20 um 20:35 schrieb Chris Wilson:
Quoting lepton (2020-07-07 19:17:51)
On Tue, Jul 7, 2020 at 10:20 AM Chris Wilson chris@chris-wilson.co.uk wrote:
Quoting lepton (2020-07-07 18:05:21)
On Tue, Jul 7, 2020 at 9:00 AM Chris Wilson chris@chris-wilson.co.uk wrote:
If we assign obj->filp, we believe that the create vgem bo is native and allow direct operations like mmap() assuming it behaves as backed by a shmemfs inode. When imported from a dmabuf, the obj->pages are not always meaningful and the shmemfs backing store misleading.
Note, that regular mmap access to a vgem bo is via the dumb buffer API, and that rejects attempts to mmap an imported dmabuf,
What do you mean by "regular mmap access" here? It looks like vgem is using vgem_gem_dumb_map as .dumb_map_offset callback then it doesn't call drm_gem_dumb_map_offset
As I too found out, and so had to correct my story telling.
By regular mmap() access I mean mmap on the vgem bo [via the dumb buffer API] as opposed to mmap() via an exported dma-buf fd. I had to look at igt to see how it was being used.
Now it seems your fix is to disable "regular mmap" on imported dma buf for vgem. I am not really a graphic guy, but then the api looks like: for a gem handle, user space has to guess to find out the way to mmap it. If user space guess wrong, then it will fail to mmap. Is this the expected way for people to handle gpu buffer?
You either have a dumb buffer handle, or a dma-buf fd. If you have the handle, you have to use the dumb buffer API, there's no other way to mmap it. If you have the dma-buf fd, you should mmap it directly. Those two are clear.
It's when you import the dma-buf into vgem and create a handle out of it, that's when the handle is no longer first class and certain uAPI [the dumb buffer API in particular] fail.
It's not brilliant, as you say, it requires the user to remember the difference between the handles, but at the same time it does prevent them falling into coherency traps by forcing them to use the right driver to handle the object, and have to consider the additional ioctls that go along with that access.
Yes, Chris is right. Mapping DMA-buf through the mmap() APIs of an importer is illegal.
What we could maybe try to do is to redirect this mmap() API call on the importer to the exporter, but I'm pretty sure that the fs layer wouldn't like that without changes.
Regards, Christian.
-Chris
On Wed, Jul 08, 2020 at 11:22:00AM +0200, Christian König wrote:
Am 07.07.20 um 20:35 schrieb Chris Wilson:
Quoting lepton (2020-07-07 19:17:51)
On Tue, Jul 7, 2020 at 10:20 AM Chris Wilson chris@chris-wilson.co.uk wrote:
Quoting lepton (2020-07-07 18:05:21)
On Tue, Jul 7, 2020 at 9:00 AM Chris Wilson chris@chris-wilson.co.uk wrote:
If we assign obj->filp, we believe that the create vgem bo is native and allow direct operations like mmap() assuming it behaves as backed by a shmemfs inode. When imported from a dmabuf, the obj->pages are not always meaningful and the shmemfs backing store misleading.
Note, that regular mmap access to a vgem bo is via the dumb buffer API, and that rejects attempts to mmap an imported dmabuf,
What do you mean by "regular mmap access" here? It looks like vgem is using vgem_gem_dumb_map as .dumb_map_offset callback then it doesn't call drm_gem_dumb_map_offset
As I too found out, and so had to correct my story telling.
By regular mmap() access I mean mmap on the vgem bo [via the dumb buffer API] as opposed to mmap() via an exported dma-buf fd. I had to look at igt to see how it was being used.
Now it seems your fix is to disable "regular mmap" on imported dma buf for vgem. I am not really a graphic guy, but then the api looks like: for a gem handle, user space has to guess to find out the way to mmap it. If user space guess wrong, then it will fail to mmap. Is this the expected way for people to handle gpu buffer?
You either have a dumb buffer handle, or a dma-buf fd. If you have the handle, you have to use the dumb buffer API, there's no other way to mmap it. If you have the dma-buf fd, you should mmap it directly. Those two are clear.
It's when you import the dma-buf into vgem and create a handle out of it, that's when the handle is no longer first class and certain uAPI [the dumb buffer API in particular] fail.
It's not brilliant, as you say, it requires the user to remember the difference between the handles, but at the same time it does prevent them falling into coherency traps by forcing them to use the right driver to handle the object, and have to consider the additional ioctls that go along with that access.
Yes, Chris is right. Mapping DMA-buf through the mmap() APIs of an importer is illegal.
What we could maybe try to do is to redirect this mmap() API call on the importer to the exporter, but I'm pretty sure that the fs layer wouldn't like that without changes.
We already do that, there's a full helper-ified path from I think shmem helpers through prime helpers to forward this all. Including handling buffer offsets and all the other lolz back&forth.
Of course there's still the problem that many drivers don't forward the cache coherency calls for begin/end cpu access, so in a bunch of cases you'll cache cacheline dirt soup. But that's kinda standard procedure for dma-buf :-P
But yeah trying to handle the mmap as an importer, bypassing the export: nope. The one exception is if you have some kind of fancy gart with cpu-visible pci bar (like at least integrated intel gpus have). But in that case the mmap very much looks&acts like device access in every way.
Cheers, Daniel
Regards, Christian.
-Chris
Am 08.07.20 um 11:54 schrieb Daniel Vetter:
On Wed, Jul 08, 2020 at 11:22:00AM +0200, Christian König wrote:
Am 07.07.20 um 20:35 schrieb Chris Wilson:
Quoting lepton (2020-07-07 19:17:51)
On Tue, Jul 7, 2020 at 10:20 AM Chris Wilson chris@chris-wilson.co.uk wrote:
Quoting lepton (2020-07-07 18:05:21)
On Tue, Jul 7, 2020 at 9:00 AM Chris Wilson chris@chris-wilson.co.uk wrote: > If we assign obj->filp, we believe that the create vgem bo is native and > allow direct operations like mmap() assuming it behaves as backed by a > shmemfs inode. When imported from a dmabuf, the obj->pages are > not always meaningful and the shmemfs backing store misleading. > > Note, that regular mmap access to a vgem bo is via the dumb buffer API, > and that rejects attempts to mmap an imported dmabuf, What do you mean by "regular mmap access" here? It looks like vgem is using vgem_gem_dumb_map as .dumb_map_offset callback then it doesn't call drm_gem_dumb_map_offset
As I too found out, and so had to correct my story telling.
By regular mmap() access I mean mmap on the vgem bo [via the dumb buffer API] as opposed to mmap() via an exported dma-buf fd. I had to look at igt to see how it was being used.
Now it seems your fix is to disable "regular mmap" on imported dma buf for vgem. I am not really a graphic guy, but then the api looks like: for a gem handle, user space has to guess to find out the way to mmap it. If user space guess wrong, then it will fail to mmap. Is this the expected way for people to handle gpu buffer?
You either have a dumb buffer handle, or a dma-buf fd. If you have the handle, you have to use the dumb buffer API, there's no other way to mmap it. If you have the dma-buf fd, you should mmap it directly. Those two are clear.
It's when you import the dma-buf into vgem and create a handle out of it, that's when the handle is no longer first class and certain uAPI [the dumb buffer API in particular] fail.
It's not brilliant, as you say, it requires the user to remember the difference between the handles, but at the same time it does prevent them falling into coherency traps by forcing them to use the right driver to handle the object, and have to consider the additional ioctls that go along with that access.
Yes, Chris is right. Mapping DMA-buf through the mmap() APIs of an importer is illegal.
What we could maybe try to do is to redirect this mmap() API call on the importer to the exporter, but I'm pretty sure that the fs layer wouldn't like that without changes.
We already do that, there's a full helper-ified path from I think shmem helpers through prime helpers to forward this all. Including handling buffer offsets and all the other lolz back&forth.
Oh, that most likely won't work correctly with unpinned DMA-bufs and needs to be avoided.
Each file descriptor is associated with an struct address_space. And when you mmap() through the importer by redirecting the system call to the exporter you end up with the wrong struct address_space in your VMA.
That in turn can go up easily in flames when the exporter tries to invalidate the CPU mappings for its DMA-buf while moving it.
Where are we doing this? My last status was that this is forbidden.
Christian.
Of course there's still the problem that many drivers don't forward the cache coherency calls for begin/end cpu access, so in a bunch of cases you'll cache cacheline dirt soup. But that's kinda standard procedure for dma-buf :-P
But yeah trying to handle the mmap as an importer, bypassing the export: nope. The one exception is if you have some kind of fancy gart with cpu-visible pci bar (like at least integrated intel gpus have). But in that case the mmap very much looks&acts like device access in every way.
Cheers, Daniel
Regards, Christian.
-Chris
On Wed, Jul 8, 2020 at 4:37 PM Christian König christian.koenig@amd.com wrote:
Am 08.07.20 um 11:54 schrieb Daniel Vetter:
On Wed, Jul 08, 2020 at 11:22:00AM +0200, Christian König wrote:
Am 07.07.20 um 20:35 schrieb Chris Wilson:
Quoting lepton (2020-07-07 19:17:51)
On Tue, Jul 7, 2020 at 10:20 AM Chris Wilson chris@chris-wilson.co.uk wrote:
Quoting lepton (2020-07-07 18:05:21) > On Tue, Jul 7, 2020 at 9:00 AM Chris Wilson chris@chris-wilson.co.uk wrote: >> If we assign obj->filp, we believe that the create vgem bo is native and >> allow direct operations like mmap() assuming it behaves as backed by a >> shmemfs inode. When imported from a dmabuf, the obj->pages are >> not always meaningful and the shmemfs backing store misleading. >> >> Note, that regular mmap access to a vgem bo is via the dumb buffer API, >> and that rejects attempts to mmap an imported dmabuf, > What do you mean by "regular mmap access" here? It looks like vgem is > using vgem_gem_dumb_map as .dumb_map_offset callback then it doesn't call > drm_gem_dumb_map_offset As I too found out, and so had to correct my story telling.
By regular mmap() access I mean mmap on the vgem bo [via the dumb buffer API] as opposed to mmap() via an exported dma-buf fd. I had to look at igt to see how it was being used.
Now it seems your fix is to disable "regular mmap" on imported dma buf for vgem. I am not really a graphic guy, but then the api looks like: for a gem handle, user space has to guess to find out the way to mmap it. If user space guess wrong, then it will fail to mmap. Is this the expected way for people to handle gpu buffer?
You either have a dumb buffer handle, or a dma-buf fd. If you have the handle, you have to use the dumb buffer API, there's no other way to mmap it. If you have the dma-buf fd, you should mmap it directly. Those two are clear.
It's when you import the dma-buf into vgem and create a handle out of it, that's when the handle is no longer first class and certain uAPI [the dumb buffer API in particular] fail.
It's not brilliant, as you say, it requires the user to remember the difference between the handles, but at the same time it does prevent them falling into coherency traps by forcing them to use the right driver to handle the object, and have to consider the additional ioctls that go along with that access.
Yes, Chris is right. Mapping DMA-buf through the mmap() APIs of an importer is illegal.
What we could maybe try to do is to redirect this mmap() API call on the importer to the exporter, but I'm pretty sure that the fs layer wouldn't like that without changes.
We already do that, there's a full helper-ified path from I think shmem helpers through prime helpers to forward this all. Including handling buffer offsets and all the other lolz back&forth.
Oh, that most likely won't work correctly with unpinned DMA-bufs and needs to be avoided.
Each file descriptor is associated with an struct address_space. And when you mmap() through the importer by redirecting the system call to the exporter you end up with the wrong struct address_space in your VMA.
That in turn can go up easily in flames when the exporter tries to invalidate the CPU mappings for its DMA-buf while moving it.
Where are we doing this? My last status was that this is forbidden.
Hm I thought we're doing all that already, but looking at the code again we're only doing this when opening a new drm fd or dma-buf fd. So the right file->f_mapping is always set at file creation time.
And we indeed don't frob this more when going another indirection ... Maybe we screwed up something somewhere :-/
Also I thought the mapping is only taken after the vma is instatiated, otherwise the tricks we're playing with dma-buf already wouldn't work: dma-buf has the buffer always at offset 0, whereas gem drm_fd mmap has it somewhere else. We already adjust vma->vm_pgoff, so I'm wondering whether we could adjust vm_file too. Or is that the thing that's forbidden? -Daniel
Christian.
Of course there's still the problem that many drivers don't forward the cache coherency calls for begin/end cpu access, so in a bunch of cases you'll cache cacheline dirt soup. But that's kinda standard procedure for dma-buf :-P
But yeah trying to handle the mmap as an importer, bypassing the export: nope. The one exception is if you have some kind of fancy gart with cpu-visible pci bar (like at least integrated intel gpus have). But in that case the mmap very much looks&acts like device access in every way.
Cheers, Daniel
Regards, Christian.
-Chris
Am 08.07.20 um 17:01 schrieb Daniel Vetter:
On Wed, Jul 8, 2020 at 4:37 PM Christian König christian.koenig@amd.com wrote:
Am 08.07.20 um 11:54 schrieb Daniel Vetter:
On Wed, Jul 08, 2020 at 11:22:00AM +0200, Christian König wrote:
Am 07.07.20 um 20:35 schrieb Chris Wilson:
Quoting lepton (2020-07-07 19:17:51)
On Tue, Jul 7, 2020 at 10:20 AM Chris Wilson chris@chris-wilson.co.uk wrote: > Quoting lepton (2020-07-07 18:05:21) >> On Tue, Jul 7, 2020 at 9:00 AM Chris Wilson chris@chris-wilson.co.uk wrote: >>> If we assign obj->filp, we believe that the create vgem bo is native and >>> allow direct operations like mmap() assuming it behaves as backed by a >>> shmemfs inode. When imported from a dmabuf, the obj->pages are >>> not always meaningful and the shmemfs backing store misleading. >>> >>> Note, that regular mmap access to a vgem bo is via the dumb buffer API, >>> and that rejects attempts to mmap an imported dmabuf, >> What do you mean by "regular mmap access" here? It looks like vgem is >> using vgem_gem_dumb_map as .dumb_map_offset callback then it doesn't call >> drm_gem_dumb_map_offset > As I too found out, and so had to correct my story telling. > > By regular mmap() access I mean mmap on the vgem bo [via the dumb buffer > API] as opposed to mmap() via an exported dma-buf fd. I had to look at > igt to see how it was being used. Now it seems your fix is to disable "regular mmap" on imported dma buf for vgem. I am not really a graphic guy, but then the api looks like: for a gem handle, user space has to guess to find out the way to mmap it. If user space guess wrong, then it will fail to mmap. Is this the expected way for people to handle gpu buffer?
You either have a dumb buffer handle, or a dma-buf fd. If you have the handle, you have to use the dumb buffer API, there's no other way to mmap it. If you have the dma-buf fd, you should mmap it directly. Those two are clear.
It's when you import the dma-buf into vgem and create a handle out of it, that's when the handle is no longer first class and certain uAPI [the dumb buffer API in particular] fail.
It's not brilliant, as you say, it requires the user to remember the difference between the handles, but at the same time it does prevent them falling into coherency traps by forcing them to use the right driver to handle the object, and have to consider the additional ioctls that go along with that access.
Yes, Chris is right. Mapping DMA-buf through the mmap() APIs of an importer is illegal.
What we could maybe try to do is to redirect this mmap() API call on the importer to the exporter, but I'm pretty sure that the fs layer wouldn't like that without changes.
We already do that, there's a full helper-ified path from I think shmem helpers through prime helpers to forward this all. Including handling buffer offsets and all the other lolz back&forth.
Oh, that most likely won't work correctly with unpinned DMA-bufs and needs to be avoided.
Each file descriptor is associated with an struct address_space. And when you mmap() through the importer by redirecting the system call to the exporter you end up with the wrong struct address_space in your VMA.
That in turn can go up easily in flames when the exporter tries to invalidate the CPU mappings for its DMA-buf while moving it.
Where are we doing this? My last status was that this is forbidden.
Hm I thought we're doing all that already, but looking at the code again we're only doing this when opening a new drm fd or dma-buf fd. So the right file->f_mapping is always set at file creation time.
And we indeed don't frob this more when going another indirection ... Maybe we screwed up something somewhere :-/
Also I thought the mapping is only taken after the vma is instatiated, otherwise the tricks we're playing with dma-buf already wouldn't work: dma-buf has the buffer always at offset 0, whereas gem drm_fd mmap has it somewhere else. We already adjust vma->vm_pgoff, so I'm wondering whether we could adjust vm_file too. Or is that the thing that's forbidden?
Yes, exactly. Modifying vm_pgoff is harmless, tons of code does that.
But changing vma->vm_file, that's something I haven't seen before and most likely could blow up badly.
Christian.
-Daniel
Christian.
Of course there's still the problem that many drivers don't forward the cache coherency calls for begin/end cpu access, so in a bunch of cases you'll cache cacheline dirt soup. But that's kinda standard procedure for dma-buf :-P
But yeah trying to handle the mmap as an importer, bypassing the export: nope. The one exception is if you have some kind of fancy gart with cpu-visible pci bar (like at least integrated intel gpus have). But in that case the mmap very much looks&acts like device access in every way.
Cheers, Daniel
Regards, Christian.
-Chris
On Wed, Jul 8, 2020 at 5:05 PM Christian König christian.koenig@amd.com wrote:
Am 08.07.20 um 17:01 schrieb Daniel Vetter:
On Wed, Jul 8, 2020 at 4:37 PM Christian König christian.koenig@amd.com wrote:
Am 08.07.20 um 11:54 schrieb Daniel Vetter:
On Wed, Jul 08, 2020 at 11:22:00AM +0200, Christian König wrote:
Am 07.07.20 um 20:35 schrieb Chris Wilson:
Quoting lepton (2020-07-07 19:17:51) > On Tue, Jul 7, 2020 at 10:20 AM Chris Wilson chris@chris-wilson.co.uk wrote: >> Quoting lepton (2020-07-07 18:05:21) >>> On Tue, Jul 7, 2020 at 9:00 AM Chris Wilson chris@chris-wilson.co.uk wrote: >>>> If we assign obj->filp, we believe that the create vgem bo is native and >>>> allow direct operations like mmap() assuming it behaves as backed by a >>>> shmemfs inode. When imported from a dmabuf, the obj->pages are >>>> not always meaningful and the shmemfs backing store misleading. >>>> >>>> Note, that regular mmap access to a vgem bo is via the dumb buffer API, >>>> and that rejects attempts to mmap an imported dmabuf, >>> What do you mean by "regular mmap access" here? It looks like vgem is >>> using vgem_gem_dumb_map as .dumb_map_offset callback then it doesn't call >>> drm_gem_dumb_map_offset >> As I too found out, and so had to correct my story telling. >> >> By regular mmap() access I mean mmap on the vgem bo [via the dumb buffer >> API] as opposed to mmap() via an exported dma-buf fd. I had to look at >> igt to see how it was being used. > Now it seems your fix is to disable "regular mmap" on imported dma buf > for vgem. I am not really a graphic guy, but then the api looks like: > for a gem handle, user space has to guess to find out the way to mmap > it. If user space guess wrong, then it will fail to mmap. Is this the > expected way > for people to handle gpu buffer? You either have a dumb buffer handle, or a dma-buf fd. If you have the handle, you have to use the dumb buffer API, there's no other way to mmap it. If you have the dma-buf fd, you should mmap it directly. Those two are clear.
It's when you import the dma-buf into vgem and create a handle out of it, that's when the handle is no longer first class and certain uAPI [the dumb buffer API in particular] fail.
It's not brilliant, as you say, it requires the user to remember the difference between the handles, but at the same time it does prevent them falling into coherency traps by forcing them to use the right driver to handle the object, and have to consider the additional ioctls that go along with that access.
Yes, Chris is right. Mapping DMA-buf through the mmap() APIs of an importer is illegal.
What we could maybe try to do is to redirect this mmap() API call on the importer to the exporter, but I'm pretty sure that the fs layer wouldn't like that without changes.
We already do that, there's a full helper-ified path from I think shmem helpers through prime helpers to forward this all. Including handling buffer offsets and all the other lolz back&forth.
Oh, that most likely won't work correctly with unpinned DMA-bufs and needs to be avoided.
Each file descriptor is associated with an struct address_space. And when you mmap() through the importer by redirecting the system call to the exporter you end up with the wrong struct address_space in your VMA.
That in turn can go up easily in flames when the exporter tries to invalidate the CPU mappings for its DMA-buf while moving it.
Where are we doing this? My last status was that this is forbidden.
Hm I thought we're doing all that already, but looking at the code again we're only doing this when opening a new drm fd or dma-buf fd. So the right file->f_mapping is always set at file creation time.
And we indeed don't frob this more when going another indirection ... Maybe we screwed up something somewhere :-/
Also I thought the mapping is only taken after the vma is instatiated, otherwise the tricks we're playing with dma-buf already wouldn't work: dma-buf has the buffer always at offset 0, whereas gem drm_fd mmap has it somewhere else. We already adjust vma->vm_pgoff, so I'm wondering whether we could adjust vm_file too. Or is that the thing that's forbidden?
Yes, exactly. Modifying vm_pgoff is harmless, tons of code does that.
But changing vma->vm_file, that's something I haven't seen before and most likely could blow up badly.
Ok, I read the shmem helpers again, I think those are the only ones which do the importer mmap -> dma_buf_mmap() forwarding, and hence break stuff all around here.
They also remove the vma->vm_pgoff offset, which means unmap_mapping_range wont work anyway. I guess for drivers which use shmem helpers the hard assumption is that a) can't use p2p dma and b) pin everything into system memory.
So not a problem. But something to keep in mind. I'll try to do a kerneldoc patch to note this somewhere. btw on that, did the timeline/syncobj documentation patch land by now? Just trying to make sure that doesn't get lost for another few months or so :-/
Cheers, Daniel
Christian.
-Daniel
Christian.
Of course there's still the problem that many drivers don't forward the cache coherency calls for begin/end cpu access, so in a bunch of cases you'll cache cacheline dirt soup. But that's kinda standard procedure for dma-buf :-P
But yeah trying to handle the mmap as an importer, bypassing the export: nope. The one exception is if you have some kind of fancy gart with cpu-visible pci bar (like at least integrated intel gpus have). But in that case the mmap very much looks&acts like device access in every way.
Cheers, Daniel
Regards, Christian.
-Chris
On Wed, Jul 8, 2020 at 6:11 PM Daniel Vetter daniel@ffwll.ch wrote:
On Wed, Jul 8, 2020 at 5:05 PM Christian König christian.koenig@amd.com wrote:
Am 08.07.20 um 17:01 schrieb Daniel Vetter:
On Wed, Jul 8, 2020 at 4:37 PM Christian König christian.koenig@amd.com wrote:
Am 08.07.20 um 11:54 schrieb Daniel Vetter:
On Wed, Jul 08, 2020 at 11:22:00AM +0200, Christian König wrote:
Am 07.07.20 um 20:35 schrieb Chris Wilson: > Quoting lepton (2020-07-07 19:17:51) >> On Tue, Jul 7, 2020 at 10:20 AM Chris Wilson chris@chris-wilson.co.uk wrote: >>> Quoting lepton (2020-07-07 18:05:21) >>>> On Tue, Jul 7, 2020 at 9:00 AM Chris Wilson chris@chris-wilson.co.uk wrote: >>>>> If we assign obj->filp, we believe that the create vgem bo is native and >>>>> allow direct operations like mmap() assuming it behaves as backed by a >>>>> shmemfs inode. When imported from a dmabuf, the obj->pages are >>>>> not always meaningful and the shmemfs backing store misleading. >>>>> >>>>> Note, that regular mmap access to a vgem bo is via the dumb buffer API, >>>>> and that rejects attempts to mmap an imported dmabuf, >>>> What do you mean by "regular mmap access" here? It looks like vgem is >>>> using vgem_gem_dumb_map as .dumb_map_offset callback then it doesn't call >>>> drm_gem_dumb_map_offset >>> As I too found out, and so had to correct my story telling. >>> >>> By regular mmap() access I mean mmap on the vgem bo [via the dumb buffer >>> API] as opposed to mmap() via an exported dma-buf fd. I had to look at >>> igt to see how it was being used. >> Now it seems your fix is to disable "regular mmap" on imported dma buf >> for vgem. I am not really a graphic guy, but then the api looks like: >> for a gem handle, user space has to guess to find out the way to mmap >> it. If user space guess wrong, then it will fail to mmap. Is this the >> expected way >> for people to handle gpu buffer? > You either have a dumb buffer handle, or a dma-buf fd. If you have the > handle, you have to use the dumb buffer API, there's no other way to > mmap it. If you have the dma-buf fd, you should mmap it directly. Those > two are clear. > > It's when you import the dma-buf into vgem and create a handle out of > it, that's when the handle is no longer first class and certain uAPI > [the dumb buffer API in particular] fail. > > It's not brilliant, as you say, it requires the user to remember the > difference between the handles, but at the same time it does prevent > them falling into coherency traps by forcing them to use the right > driver to handle the object, and have to consider the additional ioctls > that go along with that access. Yes, Chris is right. Mapping DMA-buf through the mmap() APIs of an importer is illegal.
What we could maybe try to do is to redirect this mmap() API call on the importer to the exporter, but I'm pretty sure that the fs layer wouldn't like that without changes.
We already do that, there's a full helper-ified path from I think shmem helpers through prime helpers to forward this all. Including handling buffer offsets and all the other lolz back&forth.
Oh, that most likely won't work correctly with unpinned DMA-bufs and needs to be avoided.
Each file descriptor is associated with an struct address_space. And when you mmap() through the importer by redirecting the system call to the exporter you end up with the wrong struct address_space in your VMA.
That in turn can go up easily in flames when the exporter tries to invalidate the CPU mappings for its DMA-buf while moving it.
Where are we doing this? My last status was that this is forbidden.
Hm I thought we're doing all that already, but looking at the code again we're only doing this when opening a new drm fd or dma-buf fd. So the right file->f_mapping is always set at file creation time.
And we indeed don't frob this more when going another indirection ... Maybe we screwed up something somewhere :-/
Also I thought the mapping is only taken after the vma is instatiated, otherwise the tricks we're playing with dma-buf already wouldn't work: dma-buf has the buffer always at offset 0, whereas gem drm_fd mmap has it somewhere else. We already adjust vma->vm_pgoff, so I'm wondering whether we could adjust vm_file too. Or is that the thing that's forbidden?
Yes, exactly. Modifying vm_pgoff is harmless, tons of code does that.
But changing vma->vm_file, that's something I haven't seen before and most likely could blow up badly.
Ok, I read the shmem helpers again, I think those are the only ones which do the importer mmap -> dma_buf_mmap() forwarding, and hence break stuff all around here.
They also remove the vma->vm_pgoff offset, which means unmap_mapping_range wont work anyway. I guess for drivers which use shmem helpers the hard assumption is that a) can't use p2p dma and b) pin everything into system memory.
So not a problem. But something to keep in mind. I'll try to do a kerneldoc patch to note this somewhere. btw on that, did the timeline/syncobj documentation patch land by now? Just trying to make sure that doesn't get lost for another few months or so :-/
Ok, so maybe it is a problem. Because there is a drm_gem_shmem_purge() which uses unmap_mapping_range underneath, and that's used by panfrost. And panfrost also uses the mmap helper. Kinda wondering whether we broke some stuff here, or whether the reverse map is installed before we touch vma->vm_pgoff.
panfrost folks, does panfrost purged buffer handling of mmap still work correctly? Do you have some kind of igt or similar for this?
Cheers, Daniel
Cheers, Daniel
Christian.
-Daniel
Christian.
Of course there's still the problem that many drivers don't forward the cache coherency calls for begin/end cpu access, so in a bunch of cases you'll cache cacheline dirt soup. But that's kinda standard procedure for dma-buf :-P
But yeah trying to handle the mmap as an importer, bypassing the export: nope. The one exception is if you have some kind of fancy gart with cpu-visible pci bar (like at least integrated intel gpus have). But in that case the mmap very much looks&acts like device access in every way.
Cheers, Daniel
Regards, Christian.
> -Chris
-- Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
Am 08.07.20 um 18:19 schrieb Daniel Vetter:
On Wed, Jul 8, 2020 at 6:11 PM Daniel Vetter daniel@ffwll.ch wrote:
On Wed, Jul 8, 2020 at 5:05 PM Christian König christian.koenig@amd.com wrote:
Am 08.07.20 um 17:01 schrieb Daniel Vetter:
On Wed, Jul 8, 2020 at 4:37 PM Christian König christian.koenig@amd.com wrote:
Am 08.07.20 um 11:54 schrieb Daniel Vetter:
On Wed, Jul 08, 2020 at 11:22:00AM +0200, Christian König wrote: > Am 07.07.20 um 20:35 schrieb Chris Wilson: >> Quoting lepton (2020-07-07 19:17:51) >>> On Tue, Jul 7, 2020 at 10:20 AM Chris Wilson chris@chris-wilson.co.uk wrote: >>>> Quoting lepton (2020-07-07 18:05:21) >>>>> On Tue, Jul 7, 2020 at 9:00 AM Chris Wilson chris@chris-wilson.co.uk wrote: >>>>>> If we assign obj->filp, we believe that the create vgem bo is native and >>>>>> allow direct operations like mmap() assuming it behaves as backed by a >>>>>> shmemfs inode. When imported from a dmabuf, the obj->pages are >>>>>> not always meaningful and the shmemfs backing store misleading. >>>>>> >>>>>> Note, that regular mmap access to a vgem bo is via the dumb buffer API, >>>>>> and that rejects attempts to mmap an imported dmabuf, >>>>> What do you mean by "regular mmap access" here? It looks like vgem is >>>>> using vgem_gem_dumb_map as .dumb_map_offset callback then it doesn't call >>>>> drm_gem_dumb_map_offset >>>> As I too found out, and so had to correct my story telling. >>>> >>>> By regular mmap() access I mean mmap on the vgem bo [via the dumb buffer >>>> API] as opposed to mmap() via an exported dma-buf fd. I had to look at >>>> igt to see how it was being used. >>> Now it seems your fix is to disable "regular mmap" on imported dma buf >>> for vgem. I am not really a graphic guy, but then the api looks like: >>> for a gem handle, user space has to guess to find out the way to mmap >>> it. If user space guess wrong, then it will fail to mmap. Is this the >>> expected way >>> for people to handle gpu buffer? >> You either have a dumb buffer handle, or a dma-buf fd. If you have the >> handle, you have to use the dumb buffer API, there's no other way to >> mmap it. If you have the dma-buf fd, you should mmap it directly. Those >> two are clear. >> >> It's when you import the dma-buf into vgem and create a handle out of >> it, that's when the handle is no longer first class and certain uAPI >> [the dumb buffer API in particular] fail. >> >> It's not brilliant, as you say, it requires the user to remember the >> difference between the handles, but at the same time it does prevent >> them falling into coherency traps by forcing them to use the right >> driver to handle the object, and have to consider the additional ioctls >> that go along with that access. > Yes, Chris is right. Mapping DMA-buf through the mmap() APIs of an importer > is illegal. > > What we could maybe try to do is to redirect this mmap() API call on the > importer to the exporter, but I'm pretty sure that the fs layer wouldn't > like that without changes. We already do that, there's a full helper-ified path from I think shmem helpers through prime helpers to forward this all. Including handling buffer offsets and all the other lolz back&forth.
Oh, that most likely won't work correctly with unpinned DMA-bufs and needs to be avoided.
Each file descriptor is associated with an struct address_space. And when you mmap() through the importer by redirecting the system call to the exporter you end up with the wrong struct address_space in your VMA.
That in turn can go up easily in flames when the exporter tries to invalidate the CPU mappings for its DMA-buf while moving it.
Where are we doing this? My last status was that this is forbidden.
Hm I thought we're doing all that already, but looking at the code again we're only doing this when opening a new drm fd or dma-buf fd. So the right file->f_mapping is always set at file creation time.
And we indeed don't frob this more when going another indirection ... Maybe we screwed up something somewhere :-/
Also I thought the mapping is only taken after the vma is instatiated, otherwise the tricks we're playing with dma-buf already wouldn't work: dma-buf has the buffer always at offset 0, whereas gem drm_fd mmap has it somewhere else. We already adjust vma->vm_pgoff, so I'm wondering whether we could adjust vm_file too. Or is that the thing that's forbidden?
Yes, exactly. Modifying vm_pgoff is harmless, tons of code does that.
But changing vma->vm_file, that's something I haven't seen before and most likely could blow up badly.
Ok, I read the shmem helpers again, I think those are the only ones which do the importer mmap -> dma_buf_mmap() forwarding, and hence break stuff all around here.
They also remove the vma->vm_pgoff offset, which means unmap_mapping_range wont work anyway. I guess for drivers which use shmem helpers the hard assumption is that a) can't use p2p dma and b) pin everything into system memory.
So not a problem. But something to keep in mind. I'll try to do a kerneldoc patch to note this somewhere. btw on that, did the timeline/syncobj documentation patch land by now? Just trying to make sure that doesn't get lost for another few months or so :-/
Ok, so maybe it is a problem. Because there is a drm_gem_shmem_purge() which uses unmap_mapping_range underneath, and that's used by panfrost. And panfrost also uses the mmap helper. Kinda wondering whether we broke some stuff here, or whether the reverse map is installed before we touch vma->vm_pgoff.
I think the key problem here is that unmap_mapping_range() doesn't blow up immediately when this is wrong.
E.g. we just have a stale CPU page table entry which allows userspace to write to freed up memory, but we don't really notice that immediately....
Maybe we should stop allowing to mmap() DMA-buf through the importer file descriptor altogether and only allow mapping it through its own fd or the exporter.
Christian.
panfrost folks, does panfrost purged buffer handling of mmap still work correctly? Do you have some kind of igt or similar for this?
Cheers, Daniel
Cheers, Daniel
Christian.
-Daniel
Christian.
Of course there's still the problem that many drivers don't forward the cache coherency calls for begin/end cpu access, so in a bunch of cases you'll cache cacheline dirt soup. But that's kinda standard procedure for dma-buf :-P
But yeah trying to handle the mmap as an importer, bypassing the export: nope. The one exception is if you have some kind of fancy gart with cpu-visible pci bar (like at least integrated intel gpus have). But in that case the mmap very much looks&acts like device access in every way.
Cheers, Daniel
> Regards, > Christian. > > >> -Chris
-- Daniel Vetter Software Engineer, Intel Corporation https://nam11.safelinks.protection.outlook.com/?url=http%3A%2F%2Fblog.ffwll....
On 09/07/2020 09:48, Christian König wrote:
Am 08.07.20 um 18:19 schrieb Daniel Vetter:
On Wed, Jul 8, 2020 at 6:11 PM Daniel Vetter daniel@ffwll.ch wrote:
On Wed, Jul 8, 2020 at 5:05 PM Christian König christian.koenig@amd.com wrote:
Am 08.07.20 um 17:01 schrieb Daniel Vetter:
On Wed, Jul 8, 2020 at 4:37 PM Christian König christian.koenig@amd.com wrote:
Am 08.07.20 um 11:54 schrieb Daniel Vetter: > On Wed, Jul 08, 2020 at 11:22:00AM +0200, Christian König wrote: >> Am 07.07.20 um 20:35 schrieb Chris Wilson: >>> Quoting lepton (2020-07-07 19:17:51) >>>> On Tue, Jul 7, 2020 at 10:20 AM Chris Wilson >>>> chris@chris-wilson.co.uk wrote: >>>>> Quoting lepton (2020-07-07 18:05:21) >>>>>> On Tue, Jul 7, 2020 at 9:00 AM Chris Wilson >>>>>> chris@chris-wilson.co.uk wrote: >>>>>>> If we assign obj->filp, we believe that the create vgem bo >>>>>>> is native and >>>>>>> allow direct operations like mmap() assuming it behaves as >>>>>>> backed by a >>>>>>> shmemfs inode. When imported from a dmabuf, the obj->pages are >>>>>>> not always meaningful and the shmemfs backing store >>>>>>> misleading. >>>>>>> >>>>>>> Note, that regular mmap access to a vgem bo is via the dumb >>>>>>> buffer API, >>>>>>> and that rejects attempts to mmap an imported dmabuf, >>>>>> What do you mean by "regular mmap access" here? It looks >>>>>> like vgem is >>>>>> using vgem_gem_dumb_map as .dumb_map_offset callback then it >>>>>> doesn't call >>>>>> drm_gem_dumb_map_offset >>>>> As I too found out, and so had to correct my story telling. >>>>> >>>>> By regular mmap() access I mean mmap on the vgem bo [via the >>>>> dumb buffer >>>>> API] as opposed to mmap() via an exported dma-buf fd. I had >>>>> to look at >>>>> igt to see how it was being used. >>>> Now it seems your fix is to disable "regular mmap" on imported >>>> dma buf >>>> for vgem. I am not really a graphic guy, but then the api >>>> looks like: >>>> for a gem handle, user space has to guess to find out the way >>>> to mmap >>>> it. If user space guess wrong, then it will fail to mmap. Is >>>> this the >>>> expected way >>>> for people to handle gpu buffer? >>> You either have a dumb buffer handle, or a dma-buf fd. If you >>> have the >>> handle, you have to use the dumb buffer API, there's no other >>> way to >>> mmap it. If you have the dma-buf fd, you should mmap it >>> directly. Those >>> two are clear. >>> >>> It's when you import the dma-buf into vgem and create a handle >>> out of >>> it, that's when the handle is no longer first class and certain >>> uAPI >>> [the dumb buffer API in particular] fail. >>> >>> It's not brilliant, as you say, it requires the user to >>> remember the >>> difference between the handles, but at the same time it does >>> prevent >>> them falling into coherency traps by forcing them to use the right >>> driver to handle the object, and have to consider the >>> additional ioctls >>> that go along with that access. >> Yes, Chris is right. Mapping DMA-buf through the mmap() APIs of >> an importer >> is illegal. >> >> What we could maybe try to do is to redirect this mmap() API >> call on the >> importer to the exporter, but I'm pretty sure that the fs layer >> wouldn't >> like that without changes. > We already do that, there's a full helper-ified path from I think > shmem > helpers through prime helpers to forward this all. Including > handling > buffer offsets and all the other lolz back&forth. Oh, that most likely won't work correctly with unpinned DMA-bufs and needs to be avoided.
Each file descriptor is associated with an struct address_space. And when you mmap() through the importer by redirecting the system call to the exporter you end up with the wrong struct address_space in your VMA.
That in turn can go up easily in flames when the exporter tries to invalidate the CPU mappings for its DMA-buf while moving it.
Where are we doing this? My last status was that this is forbidden.
Hm I thought we're doing all that already, but looking at the code again we're only doing this when opening a new drm fd or dma-buf fd. So the right file->f_mapping is always set at file creation time.
And we indeed don't frob this more when going another indirection ... Maybe we screwed up something somewhere :-/
Also I thought the mapping is only taken after the vma is instatiated, otherwise the tricks we're playing with dma-buf already wouldn't work: dma-buf has the buffer always at offset 0, whereas gem drm_fd mmap has it somewhere else. We already adjust vma->vm_pgoff, so I'm wondering whether we could adjust vm_file too. Or is that the thing that's forbidden?
Yes, exactly. Modifying vm_pgoff is harmless, tons of code does that.
But changing vma->vm_file, that's something I haven't seen before and most likely could blow up badly.
Ok, I read the shmem helpers again, I think those are the only ones which do the importer mmap -> dma_buf_mmap() forwarding, and hence break stuff all around here.
They also remove the vma->vm_pgoff offset, which means unmap_mapping_range wont work anyway. I guess for drivers which use shmem helpers the hard assumption is that a) can't use p2p dma and b) pin everything into system memory.
So not a problem. But something to keep in mind. I'll try to do a kerneldoc patch to note this somewhere. btw on that, did the timeline/syncobj documentation patch land by now? Just trying to make sure that doesn't get lost for another few months or so :-/
Ok, so maybe it is a problem. Because there is a drm_gem_shmem_purge() which uses unmap_mapping_range underneath, and that's used by panfrost. And panfrost also uses the mmap helper. Kinda wondering whether we broke some stuff here, or whether the reverse map is installed before we touch vma->vm_pgoff.
I think the key problem here is that unmap_mapping_range() doesn't blow up immediately when this is wrong.
E.g. we just have a stale CPU page table entry which allows userspace to write to freed up memory, but we don't really notice that immediately....
Maybe we should stop allowing to mmap() DMA-buf through the importer file descriptor altogether and only allow mapping it through its own fd or the exporter.
That is what I tried to do with panfrost a while ago:
583bbf46133c drm/panfrost: Use drm_gem_map_offset()
panfrost_ioctl_mmap_bo() contains a reimplementation of drm_gem_map_offset() but with a bug - it allows mapping imported objects (without going through the exporter). Fix this by switching to use the newly renamed drm_gem_map_offset() function instead which has the bonus of simplifying the code.
Sadly it was followed by:
be855382bacb Revert "drm/panfrost: Use drm_gem_map_offset()" This reverts commit 583bbf46133c726bae277e8f4e32bfba2a528c7f.
Turns out we need mmap to work on imported BOs even if the current code is buggy.
Christian.
panfrost folks, does panfrost purged buffer handling of mmap still work correctly? Do you have some kind of igt or similar for this?
I'm not aware of any real testing of this. And I fear it probably isn't getting much in the way of real-world testing either otherwise someone would have grown tired of the "Purging xx bytes" message[1]
I'm a little bit lost on this thread - are you expecting this patch to break panfrost? We shouldn't be purging imported memory. Although I'm not sure what (if anything) stops you trying to mark imported memory as "don't need". Or indeed what would happen.
Steve
[1] I have a patch silencing that, but recently haven't had much time for working on Panfrost, and don't have my WFH setup quite as slick as I did when I was in the office.
Am 09.07.20 um 15:54 schrieb Steven Price:
On 09/07/2020 09:48, Christian König wrote:
Am 08.07.20 um 18:19 schrieb Daniel Vetter:
On Wed, Jul 8, 2020 at 6:11 PM Daniel Vetter daniel@ffwll.ch wrote:
On Wed, Jul 8, 2020 at 5:05 PM Christian König christian.koenig@amd.com wrote:
Am 08.07.20 um 17:01 schrieb Daniel Vetter:
On Wed, Jul 8, 2020 at 4:37 PM Christian König christian.koenig@amd.com wrote: > Am 08.07.20 um 11:54 schrieb Daniel Vetter: >> On Wed, Jul 08, 2020 at 11:22:00AM +0200, Christian König wrote: >>> Am 07.07.20 um 20:35 schrieb Chris Wilson: >>>> Quoting lepton (2020-07-07 19:17:51) >>>>> On Tue, Jul 7, 2020 at 10:20 AM Chris Wilson >>>>> chris@chris-wilson.co.uk wrote: >>>>>> Quoting lepton (2020-07-07 18:05:21) >>>>>>> On Tue, Jul 7, 2020 at 9:00 AM Chris Wilson >>>>>>> chris@chris-wilson.co.uk wrote: >>>>>>>> If we assign obj->filp, we believe that the create vgem >>>>>>>> bo is native and >>>>>>>> allow direct operations like mmap() assuming it behaves >>>>>>>> as backed by a >>>>>>>> shmemfs inode. When imported from a dmabuf, the >>>>>>>> obj->pages are >>>>>>>> not always meaningful and the shmemfs backing store >>>>>>>> misleading. >>>>>>>> >>>>>>>> Note, that regular mmap access to a vgem bo is via the >>>>>>>> dumb buffer API, >>>>>>>> and that rejects attempts to mmap an imported dmabuf, >>>>>>> What do you mean by "regular mmap access" here? It looks >>>>>>> like vgem is >>>>>>> using vgem_gem_dumb_map as .dumb_map_offset callback then >>>>>>> it doesn't call >>>>>>> drm_gem_dumb_map_offset >>>>>> As I too found out, and so had to correct my story telling. >>>>>> >>>>>> By regular mmap() access I mean mmap on the vgem bo [via >>>>>> the dumb buffer >>>>>> API] as opposed to mmap() via an exported dma-buf fd. I had >>>>>> to look at >>>>>> igt to see how it was being used. >>>>> Now it seems your fix is to disable "regular mmap" on >>>>> imported dma buf >>>>> for vgem. I am not really a graphic guy, but then the api >>>>> looks like: >>>>> for a gem handle, user space has to guess to find out the >>>>> way to mmap >>>>> it. If user space guess wrong, then it will fail to mmap. Is >>>>> this the >>>>> expected way >>>>> for people to handle gpu buffer? >>>> You either have a dumb buffer handle, or a dma-buf fd. If you >>>> have the >>>> handle, you have to use the dumb buffer API, there's no other >>>> way to >>>> mmap it. If you have the dma-buf fd, you should mmap it >>>> directly. Those >>>> two are clear. >>>> >>>> It's when you import the dma-buf into vgem and create a >>>> handle out of >>>> it, that's when the handle is no longer first class and >>>> certain uAPI >>>> [the dumb buffer API in particular] fail. >>>> >>>> It's not brilliant, as you say, it requires the user to >>>> remember the >>>> difference between the handles, but at the same time it does >>>> prevent >>>> them falling into coherency traps by forcing them to use the >>>> right >>>> driver to handle the object, and have to consider the >>>> additional ioctls >>>> that go along with that access. >>> Yes, Chris is right. Mapping DMA-buf through the mmap() APIs >>> of an importer >>> is illegal. >>> >>> What we could maybe try to do is to redirect this mmap() API >>> call on the >>> importer to the exporter, but I'm pretty sure that the fs >>> layer wouldn't >>> like that without changes. >> We already do that, there's a full helper-ified path from I >> think shmem >> helpers through prime helpers to forward this all. Including >> handling >> buffer offsets and all the other lolz back&forth. > Oh, that most likely won't work correctly with unpinned DMA-bufs > and > needs to be avoided. > > Each file descriptor is associated with an struct address_space. > And > when you mmap() through the importer by redirecting the system > call to > the exporter you end up with the wrong struct address_space in > your VMA. > > That in turn can go up easily in flames when the exporter tries to > invalidate the CPU mappings for its DMA-buf while moving it. > > Where are we doing this? My last status was that this is forbidden. Hm I thought we're doing all that already, but looking at the code again we're only doing this when opening a new drm fd or dma-buf fd. So the right file->f_mapping is always set at file creation time.
And we indeed don't frob this more when going another indirection ... Maybe we screwed up something somewhere :-/
Also I thought the mapping is only taken after the vma is instatiated, otherwise the tricks we're playing with dma-buf already wouldn't work: dma-buf has the buffer always at offset 0, whereas gem drm_fd mmap has it somewhere else. We already adjust vma->vm_pgoff, so I'm wondering whether we could adjust vm_file too. Or is that the thing that's forbidden?
Yes, exactly. Modifying vm_pgoff is harmless, tons of code does that.
But changing vma->vm_file, that's something I haven't seen before and most likely could blow up badly.
Ok, I read the shmem helpers again, I think those are the only ones which do the importer mmap -> dma_buf_mmap() forwarding, and hence break stuff all around here.
They also remove the vma->vm_pgoff offset, which means unmap_mapping_range wont work anyway. I guess for drivers which use shmem helpers the hard assumption is that a) can't use p2p dma and b) pin everything into system memory.
So not a problem. But something to keep in mind. I'll try to do a kerneldoc patch to note this somewhere. btw on that, did the timeline/syncobj documentation patch land by now? Just trying to make sure that doesn't get lost for another few months or so :-/
Ok, so maybe it is a problem. Because there is a drm_gem_shmem_purge() which uses unmap_mapping_range underneath, and that's used by panfrost. And panfrost also uses the mmap helper. Kinda wondering whether we broke some stuff here, or whether the reverse map is installed before we touch vma->vm_pgoff.
I think the key problem here is that unmap_mapping_range() doesn't blow up immediately when this is wrong.
E.g. we just have a stale CPU page table entry which allows userspace to write to freed up memory, but we don't really notice that immediately....
Maybe we should stop allowing to mmap() DMA-buf through the importer file descriptor altogether and only allow mapping it through its own fd or the exporter.
That is what I tried to do with panfrost a while ago:
583bbf46133c drm/panfrost: Use drm_gem_map_offset()
panfrost_ioctl_mmap_bo() contains a reimplementation of drm_gem_map_offset() but with a bug - it allows mapping imported objects (without going through the exporter). Fix this by switching to use the newly renamed drm_gem_map_offset() function instead which has the bonus of simplifying the code.
Sadly it was followed by:
be855382bacb Revert "drm/panfrost: Use drm_gem_map_offset()" This reverts commit 583bbf46133c726bae277e8f4e32bfba2a528c7f.
Turns out we need mmap to work on imported BOs even if the current code is buggy.
Mhm, we might need to push back on the revert.
Christian.
panfrost folks, does panfrost purged buffer handling of mmap still work correctly? Do you have some kind of igt or similar for this?
I'm not aware of any real testing of this. And I fear it probably isn't getting much in the way of real-world testing either otherwise someone would have grown tired of the "Purging xx bytes" message[1]
I'm a little bit lost on this thread - are you expecting this patch to break panfrost? We shouldn't be purging imported memory. Although I'm not sure what (if anything) stops you trying to mark imported memory as "don't need". Or indeed what would happen.
As long as panfrost keeps the imported DMA-bufs pinned everything should be fine as it is.
But this can open up a security hole you can push an elephant through. So you need to keep this in the back of your mind if this is ever implemented.
Christian.
Steve
[1] I have a patch silencing that, but recently haven't had much time for working on Panfrost, and don't have my WFH setup quite as slick as I did when I was in the office.
Am 08.07.20 um 18:11 schrieb Daniel Vetter:
On Wed, Jul 8, 2020 at 5:05 PM Christian König christian.koenig@amd.com wrote:
Am 08.07.20 um 17:01 schrieb Daniel Vetter:
On Wed, Jul 8, 2020 at 4:37 PM Christian König christian.koenig@amd.com wrote:
Am 08.07.20 um 11:54 schrieb Daniel Vetter:
On Wed, Jul 08, 2020 at 11:22:00AM +0200, Christian König wrote:
Am 07.07.20 um 20:35 schrieb Chris Wilson: > Quoting lepton (2020-07-07 19:17:51) >> On Tue, Jul 7, 2020 at 10:20 AM Chris Wilson chris@chris-wilson.co.uk wrote: >>> Quoting lepton (2020-07-07 18:05:21) >>>> On Tue, Jul 7, 2020 at 9:00 AM Chris Wilson chris@chris-wilson.co.uk wrote: >>>>> If we assign obj->filp, we believe that the create vgem bo is native and >>>>> allow direct operations like mmap() assuming it behaves as backed by a >>>>> shmemfs inode. When imported from a dmabuf, the obj->pages are >>>>> not always meaningful and the shmemfs backing store misleading. >>>>> >>>>> Note, that regular mmap access to a vgem bo is via the dumb buffer API, >>>>> and that rejects attempts to mmap an imported dmabuf, >>>> What do you mean by "regular mmap access" here? It looks like vgem is >>>> using vgem_gem_dumb_map as .dumb_map_offset callback then it doesn't call >>>> drm_gem_dumb_map_offset >>> As I too found out, and so had to correct my story telling. >>> >>> By regular mmap() access I mean mmap on the vgem bo [via the dumb buffer >>> API] as opposed to mmap() via an exported dma-buf fd. I had to look at >>> igt to see how it was being used. >> Now it seems your fix is to disable "regular mmap" on imported dma buf >> for vgem. I am not really a graphic guy, but then the api looks like: >> for a gem handle, user space has to guess to find out the way to mmap >> it. If user space guess wrong, then it will fail to mmap. Is this the >> expected way >> for people to handle gpu buffer? > You either have a dumb buffer handle, or a dma-buf fd. If you have the > handle, you have to use the dumb buffer API, there's no other way to > mmap it. If you have the dma-buf fd, you should mmap it directly. Those > two are clear. > > It's when you import the dma-buf into vgem and create a handle out of > it, that's when the handle is no longer first class and certain uAPI > [the dumb buffer API in particular] fail. > > It's not brilliant, as you say, it requires the user to remember the > difference between the handles, but at the same time it does prevent > them falling into coherency traps by forcing them to use the right > driver to handle the object, and have to consider the additional ioctls > that go along with that access. Yes, Chris is right. Mapping DMA-buf through the mmap() APIs of an importer is illegal.
What we could maybe try to do is to redirect this mmap() API call on the importer to the exporter, but I'm pretty sure that the fs layer wouldn't like that without changes.
We already do that, there's a full helper-ified path from I think shmem helpers through prime helpers to forward this all. Including handling buffer offsets and all the other lolz back&forth.
Oh, that most likely won't work correctly with unpinned DMA-bufs and needs to be avoided.
Each file descriptor is associated with an struct address_space. And when you mmap() through the importer by redirecting the system call to the exporter you end up with the wrong struct address_space in your VMA.
That in turn can go up easily in flames when the exporter tries to invalidate the CPU mappings for its DMA-buf while moving it.
Where are we doing this? My last status was that this is forbidden.
Hm I thought we're doing all that already, but looking at the code again we're only doing this when opening a new drm fd or dma-buf fd. So the right file->f_mapping is always set at file creation time.
And we indeed don't frob this more when going another indirection ... Maybe we screwed up something somewhere :-/
Also I thought the mapping is only taken after the vma is instatiated, otherwise the tricks we're playing with dma-buf already wouldn't work: dma-buf has the buffer always at offset 0, whereas gem drm_fd mmap has it somewhere else. We already adjust vma->vm_pgoff, so I'm wondering whether we could adjust vm_file too. Or is that the thing that's forbidden?
Yes, exactly. Modifying vm_pgoff is harmless, tons of code does that.
But changing vma->vm_file, that's something I haven't seen before and most likely could blow up badly.
Ok, I read the shmem helpers again, I think those are the only ones which do the importer mmap -> dma_buf_mmap() forwarding, and hence break stuff all around here.
They also remove the vma->vm_pgoff offset, which means unmap_mapping_range wont work anyway. I guess for drivers which use shmem helpers the hard assumption is that a) can't use p2p dma and b) pin everything into system memory.
So not a problem. But something to keep in mind. I'll try to do a kerneldoc patch to note this somewhere. btw on that, did the timeline/syncobj documentation patch land by now? Just trying to make sure that doesn't get lost for another few months or so :-/
I still haven't found time to double check the documentation and most likely won't in quite a while.
Sorry for that, but yeah you know :)
Christian.
Cheers, Daniel
Christian.
-Daniel
Christian.
Of course there's still the problem that many drivers don't forward the cache coherency calls for begin/end cpu access, so in a bunch of cases you'll cache cacheline dirt soup. But that's kinda standard procedure for dma-buf :-P
But yeah trying to handle the mmap as an importer, bypassing the export: nope. The one exception is if you have some kind of fancy gart with cpu-visible pci bar (like at least integrated intel gpus have). But in that case the mmap very much looks&acts like device access in every way.
Cheers, Daniel
Regards, Christian.
> -Chris
linux-stable-mirror@lists.linaro.org
-
Christian König -
Daniel Vetter -
Steven Price