Re: [Regression] drm/scheduler: track GPU active time per entity
From: Asahi Lina
Date: Thu Apr 06 2023 - 10:45:14 EST
On 06/04/2023 22.37, Daniel Vetter wrote:
On Thu, Apr 06, 2023 at 09:21:47PM +0900, Asahi Lina wrote:
On 06/04/2023 19.09, Daniel Vetter wrote:
On Thu, Apr 06, 2023 at 06:05:11PM +0900, Asahi Lina wrote:
On 06/04/2023 17.27, Daniel Vetter wrote:
On Thu, 6 Apr 2023 at 10:22, Christian König <christian.koenig@xxxxxxx> wrote:
Am 05.04.23 um 18:09 schrieb Luben Tuikov:
On 2023-04-05 10:05, Danilo Krummrich wrote:
On 4/4/23 06:31, Luben Tuikov wrote:
On 2023-03-28 04:54, Lucas Stach wrote:
Hi Danilo,
Am Dienstag, dem 28.03.2023 um 02:57 +0200 schrieb Danilo Krummrich:
Hi all,
Commit df622729ddbf ("drm/scheduler: track GPU active time per entity")
tries to track the accumulated time that a job was active on the GPU
writing it to the entity through which the job was deployed to the
scheduler originally. This is done within drm_sched_get_cleanup_job()
which fetches a job from the schedulers pending_list.
Doing this can result in a race condition where the entity is already
freed, but the entity's newly added elapsed_ns field is still accessed
once the job is fetched from the pending_list.
After drm_sched_entity_destroy() being called it should be safe to free
the structure that embeds the entity. However, a job originally handed
over to the scheduler by this entity might still reside in the
schedulers pending_list for cleanup after drm_sched_entity_destroy()
already being called and the entity being freed. Hence, we can run into
a UAF.
Sorry about that, I clearly didn't properly consider this case.
In my case it happened that a job, as explained above, was just picked
from the schedulers pending_list after the entity was freed due to the
client application exiting. Meanwhile this freed up memory was already
allocated for a subsequent client applications job structure again.
Hence, the new jobs memory got corrupted. Luckily, I was able to
reproduce the same corruption over and over again by just using
deqp-runner to run a specific set of VK test cases in parallel.
Fixing this issue doesn't seem to be very straightforward though (unless
I miss something), which is why I'm writing this mail instead of sending
a fix directly.
Spontaneously, I see three options to fix it:
1. Rather than embedding the entity into driver specific structures
(e.g. tied to file_priv) we could allocate the entity separately and
reference count it, such that it's only freed up once all jobs that were
deployed through this entity are fetched from the schedulers pending list.
My vote is on this or something in similar vain for the long term. I
have some hope to be able to add a GPU scheduling algorithm with a bit
more fairness than the current one sometime in the future, which
requires execution time tracking on the entities.
Danilo,
Using kref is preferable, i.e. option 1 above.
I think the only real motivation for doing that would be for generically
tracking job statistics within the entity a job was deployed through. If
we all agree on tracking job statistics this way I am happy to prepare a
patch for this option and drop this one:
https://lore.kernel.org/all/20230331000622.4156-1-dakr@xxxxxxxxxx/T/#u
Hmm, I never thought about "job statistics" when I preferred using kref above.
The reason kref is attractive is because one doesn't need to worry about
it--when the last user drops the kref, the release is called to do
housekeeping. If this never happens, we know that we have a bug to debug.
Yeah, reference counting unfortunately have some traps as well. For
example rarely dropping the last reference from interrupt context or
with some unexpected locks help when the cleanup function doesn't expect
that is a good recipe for problems as well.
Regarding the patch above--I did look around the code, and it seems safe,
as per your analysis, I didn't see any reference to entity after job submission,
but I'll comment on that thread as well for the record.
Reference counting the entities was suggested before. The intentionally
avoided that so far because the entity might be the tip of the iceberg
of stuff you need to keep around.
For example for command submission you also need the VM and when you
keep the VM alive you also need to keep the file private alive....
Yeah refcounting looks often like the easy way out to avoid
use-after-free issue, until you realize you've just made lifetimes
unbounded and have some enourmous leaks: entity keeps vm alive, vm
keeps all the bo alives, somehow every crash wastes more memory
because vk_device_lost means userspace allocates new stuff for
everything.
Refcounting everywhere has been working well for us, so well that so far all
the oopses we've hit have been... drm_sched bugs like this one, not anything
in the driver. But at least in Rust you have the advantage that you can't
just forget a decref in a rarely-hit error path (or worse, forget an incref
somewhere important)... ^^
If possible a lifetime design where lifetimes have hard bounds and you
just borrow a reference under a lock (or some other ownership rule) is
generally much more solid. But also much harder to design correctly
:-/
Additional to that we have some ugly inter dependencies between tearing
down an application (potential with a KILL signal from the OOM killer)
and backward compatibility for some applications which render something
and quit before the rendering is completed in the hardware.
Yeah I think that part would also be good to sort out once&for all in
drm/sched, because i915 has/had the same struggle.
-Daniel
Is this really a thing? I think that's never going to work well for explicit
sync, since the kernel doesn't even know what BOs it has to keep alive for a
job... I guess it could keep the entire file and all of its objects/VMs/etc
alive until all of its submissions complete but... ewww.
Our Mesa implementation synchronously waits for all jobs on context destroy
for this reason, but if you just kill the app, yeah, you get faults as
running GPU jobs have BOs yanked out from under them. Kill loops make for a
good way of testing fault handling...
You wind down the entire thing on file close? Like
- stop all context
- tear down all context
- tear down all vm
- tear down all obj
Just winding things down in a random order and then letting gpu fault
handling sort out the mess doesn't strike me as particularly clean design
...
The idea is that object drop order generally doesn't matter since things
that care about other things should own them or hold references to them
anyway, so the dependency graph of all the resources is encoded directly in
the type hierarchy instead of having to open-code a "cleanup procedure"...
which then invariably leads to corner cases when you have to do the same
thing, or part of it, for error handling.
This has been working *very* well! It solves the issue of error handling
since error handling just unwinds whatever was done to that point naturally
in Rust (? operator), so there's practically no open-coded error handling
code anywhere. The first time we ran into OOMs (Xonotic with no Mesa texture
compression support yet, on 8GB machines on max settings...) the whole thing
just worked. OOM killer went rampant and shmem doesn't account stuff to
processes properly of course, but all the error paths, allocation errors,
etc... all of that just worked, first try, across dozens of error paths that
had never been tested before, not a single oops or deadlock or leak or
anything in sight. Similarly, yesterday I did manage to run into drm_sched
failing to create kthreads (the scaling issue Matthew's series fixes)... and
still, that was fine. That happens on queue creation so it just bubbled up
to Mesa as a failed ioctl and things kept moving along nicely otherwise. I
even have nice ergonomic XArray semantics so that you can reserve a new
slot, allocate some object, then populate it, and if you don't (because you
errored out in between) it automatically gets freed again without explicit
cleanup code.
And it also means that I can encode *firmware* resource dependencies in the
type system (with Rust lifetimes attached to *GPU* pointers even - it's a
bit dodgy how it's done but it works well in practice). Since it is
absolutely critical that the firmware objects respect their lifetimes or
else the whole thing crashes irrecoverably, this is the only way I feel it's
been even practical to write this driver and not be a firmware crash mess.
Of course we still get some crashes due to flaws in how I understand the
firmware, but it's always things I don't know, not things I accidentally
messed up in some corner case code path we don't normally hit, since I just
don't have to think about that as long as the hierarchy is right.
I actually don't know exactly what precise order things get dropped in for
this reason! I could find out, and it's predictable in Rust, what I mean is
that thinking about a total order like that is not necessary for correctness
as long as I got the ownership right. Off the top of my head though, it goes
very roughly like this:
- On File close, all the GEM objects get closed (DRM core does this)
- This triggers explicit drops of all mappings in those GEM objects owned by
that File (identified by unique ID, this is the one annoying circular
reference thing I mentioned in the other thread...). At this point the GPU
probably faults but we don't care. *
- The File itself gets dropped, which drops the XArrays for queues and
(UAPI) VMs
- UAPI VMs getting dropped doesn't do much other than unmap a single dummy
object. The underlying MMU VM is refcounted and jobs hold references. This
also drops the userspace VM object allocator used for kernel-managed
allocations, but that too is internally refcounted and won't go away until
all of its allocations do.
- Queues get dropped, which mostly drops a bunch of references to things
that no longer matter, along with the scheduler and scheduler entity.
- The entity already has a reference to the scheduler in the abstraction (to
meet the soundness requirement), so the entity necessarily goes first. That
kills all not yet scheduled jobs, freeing any resources they might use.
- Then the scheduler gets torn down, and with my other patch that logically
kills all in-flight jobs, detaching their hardware fences and dropping the
job objects. This... still doesn't do much other than drop some references
that we don't care about.
- At this point, if any jobs are in flight, their firmware objects and all
of the type hierarchy that goes with them is still alive, as well as the
firmware queues they're in and the Rust objects representing them, the VMs
they use, the Events they have been allocated...
- Jobs complete (successfully or fault), then when complete get popped off
of the Rust-side queue objects that represent the firmware queues.
- When any given FW queue is empty, it relinquishes its assigned firmware
event ID. That causes the event system to drop its owner reference to it,
which means the queue itself gets dropped (since the UAPI Queue that also
held a reference is gone). That then also drops a reference to what I call
the GpuContext.
- To avoid deadlocks, completed job objects are freed in another thread
(ugly hack right now, should be done better in the future). Eventually as
that happens, any resources they reference are dropped, including some
shared ones that are logically tied to the scheduler/queues, references to
the MMU VM address space, references to the VM slot that address space is
assigned to, objects allocated out of user VM space, everything. Refcounts
everywhere for anything shared, direct ownership of child structures for
anything that isn't (work buffers, firmware command lists, etc.). I once
tried to make a slide of the references and pointers involved in just the
vertex half of a single GPU job and... even just that was quite interesting.
- When the last job completes, that drops the last reference to the VM slot,
which means the backing VM is logically detached from the GPU MMU (this is
lazy though, it's still there in practice).
- When the last firmware queue is dropped for a logical queue/sched/etc
(which means no more jobs are running at the GPU for that context), that
drops the last reference to the GpuContext. That also gets shoved into
another thread context for cleanup to avoid deadlocks with fault recovery.
- When that is finally cleaned up, a firmware command is sent to invalidate
the GpuContext. I'm still figuring out what that does and what the lifetime
rules are here (this is the only explicit invalidation command that exists),
but as of yesterday I know that at the very least we need to keep hold of
any Tiled Vertex Buffer associated with it until after inval, so that now
has a reference to it that gets dropped after the firmware acknowledges the
invalidation (unless it's a non-render-capable Queue, then no TVB
necessary).
- When the Buffer gets dropped, that frees both its backing memory and
(virtual) page list structures, which are in user VM space, as well as some
kernel firmware objects.
- If things have happened in the order I listed here, those will be the last
allocations in the two user VM space heap object allocators, so those now
get dropped, which drops the mappings of their backing GEM objects,
unmapping them from the MMU VM page tables.
- Those mappings will now be the last references to the actual MMU VM
object, so that it gets destroyed (the lazy detach comes into effect here,
PT base address is removed from the VM context base table, full ASID
invalidate, etc.), which with it drops the IoPgTable that backs it, which
frees the page tables.
- Then finally the GEM objects backing the userspace allocators get dropped
as well, which will be the last reference to them, so those get freed.
Thanks for the write up!
Maybe some more fundamental thoughts on this entire endeavour:
I think one thing that's causing a lot of friction that in C drivers at
least some of these things are done with implied/borrowed references. If
you want an absolute shocking example, the gpu crash dump sampling for a
driver that does dynamic memory management can be pretty epic:
- the timeout handler is guaranteed (well if the driver didn't screw up
things, which mostly boils down to call drm_sched_stop() before it
analyzes anything at all wrt gpu hw state) to hold up the drm_job
completion fence
- this fence is (at submit ioctl time) installed into all the dma_resv
fence containers for the vm (single shared dma_resv for all vm-private obj for
efficiency, all shareable gem_bo have their own dma_resv)
- the memory manager guarantees that it will not free (or even move in the
case of ttm) the buffer while a fence is unsingalled. ttm does this by
delaying the free as needed, i915-gem and some of the others did this by
trying to hold onto extra references. the latter causes way too much
problems with reference dropping in my experience looking at a decade of
drivers getting details wrong here.
- which means the crash dump handler can have a plain list of gem_bo to
record, without any references, because they wont go away before it
finishes.
- more fun even, you could directly sample the memory locations at ioctl
submit time, and even when ttm moves the buffers around in a pipelined
fashion: your timeout handler wont run before all the jobs to move the
buffer into the right location have completed, and it will not unblock
any subsequent move or eviction that's already queued up. Which means
even the raw memory address will point at the right stuff.
This is just one example. drm and the kernel is full of these borrowed
reference tricks with often cross through the entire stack, so rust has to
be able to cope. The sales pitch of rust, and really the reason it's the
first non-C language in linux, is that with the borrow semantics, it can
validate this stuff at compile time, and allow kernel drivers to continue
playing these tricks to outright avoid any and all refcounting needs. If
rust doesn't deliver and we need to have all the refcounting again to make
rust drivers safe I think that'll cast serious doubts on the viability of
rust-in-linux.
Right, so the thing is Rust can encode *some* borrowed reference
semantics at compile time, namely what Rust lifetimes can represent. But
you can't encode arbitrarily complex "I can guarantee this is alive
right now because reasons A,B,C,D" semantics in the type system. For
arbitrarily complex rules like that, you need to either refcount, or use
unsafe and raw pointers. For example, in my driver I use raw pointers in
the event slot acquisition to point to the actual u32s that hold the
event values, because I can guarantee through the semantics of that
module that those will always be valid pointers to the live array (and
never alias too), even though I can't encode that in the type system.
Throwing around refcounted pointers to the whole object there would be
silly, so instead I just used unsafe and left a SAFETY comment
explaining why it's safe.
In the end, unsafe isn't any worse than C (it's strictly better, you
still get the borrow checker and type safety and everything else), so I
don't think Rust not being able to solve all of these problems is really
a good argument against Rust. The idea, and where Rust really shines, is
that even when you have to do this you are *containing* the unsafe code
within places where it can be audited and explained. And when the
majority of the code is safe, and when you strive to make the interfaces
between modules safe, and when you limit the ability of cross-module
interactions to break safety (Not eliminate! In abstractions yes, and in
userspace Rust this is expected, but within a driver it gets more
interesting and I can spend a long time talking about how safety
boundaries in drivers are complicated and a bit porous...) you end up
reducing the chances of safety bugs by orders of magnitude anyway.
And it worked, right? I have ~128 unsafe{} blocks in the driver and
still no oopses in production ^^
But (big but!) the idea with safe Rust abstractions (the stuff in rust/)
is that they absolutely do need to be safe (unless marked unsafe, but
that sucks...), because then implicitly you are documenting the
safety/lifetime requirements in the type system and that's hugely
beneficial for common code. It means users of those APIs don't have to
think about the internals and safety invariants and all that. Drivers
still have to worry about internal safety, but they don't have to worry
about messing up the interface with common code. That's also very
beneficial for allowing refactoring of that common code without worrying
that you'll break users.
So technically we could just give up and mark the drm_sched abstraction
unsafe and actually document all of these safety requirements (which is
the other side of the coin - if you want to use unsafe, you get to write
the docs, lawyer style! That's the rule! No getting by with vague docs
with dozens of unexplained corner cases!). But, honestly, I think that
would be kind of sad... drm_sched really should have a self-contained,
safe API, and I just don't see a good reason why it can't other than it
not having been designed with this in mind initially. It was ported out
of a driver, and not that long ago, right? That explains a lot, because
it means it is probably leaking all of these assumptions from that
driver's internal design... and it probably doesn't help that the people
maintaining it all seem to be from that particular company either. It's
easy to have tunnel vision if you mostly work on one use case...
Now there's definitely going to be hilarious bugs uncovered on the C side,
and semantics that need to be clarified, but I think enabling scary tricks
like the above one is what'll fundamentally make or break rust in linux.
For the specific case of crash dumps and stop-the-world stuff like
that... yeah, those are some of the worst things to fit in otherwise
nice Rust designs because they violate layering left and right. I don't
have a good answer for how exactly I'd do this in my driver yet. I'm
sure it *can* be done, and there's always unsafe if needed, since at
that point you're dealing in driver code and that's a lot less
contentious than an abstraction being unsafe itself...
My initial plan for FW crash dumps is going to be fairly easy because I
only really care about dumping the firmware object arena and maybe
actual firmware .data, and that's all GPU-global anyway, and it won't be
hard to poke my way into the global allocators to take a snapshot of
their backing GEM objects. There's a global alloc lock involved there
which is all I need for basic consistency and getting the locking right
for allocs as far as deadlock risk is something I need to care about for
other reasons anyway. It won't guarantee that nobody else is *writing*
to any of that memory while I take a snapshot, but I don't really care
about that since it would be rare and should be limited to
newly-allocated memory the firmware doesn't yet care about anyway.
In the end, nothing stops you from having piles of raw pointer
backdoors, having a rule that "this is only used for the crash handler",
and then having a couple pages of SAFETY comments in that bit of the
code with a big "here be dragons" disclaimer. It won't make that
particular chunk of code any less likely to have bugs than the
equivalent C code, but at least the rest of the driver isn't affected,
and hopefully you can at least avoid unsafe in some subsets of the
troublesome part too ^^
There's also a flipside to this though: Who cares if you take extra
redundant GEM BO references in the crash handler due to API safety?
After all, it's not a perf-sensitive codepath. Of course, if safety
causes you to add refcounting at all where none would otherwise be
needed, or to have to do more of it in hot paths, or to introduce extra
locking that could cause deadlocks that's a different story. But I think
the thought process is very different between Rust and C here, when
writing drivers. In C, it takes extra work to be safe, so chances are
most people won't write defensive code like that even where there is no
performance reason not to, since it's outright easier to read code the
less safe it is. Rust isn't like that! Most safety is encoded in types
(automatic reference counting is a big one here), so it doesn't take any
extra effort to be safe, while it usually does to do it unsafely (unsafe
{} blocks if nothing else). So then you write safe code by default...
which yes, might do some extra refcounting and locking and stuff, but
you can always optimize that later, profiler in hand. Once you know you
need unsafe for real perf reasons, by all means, and I expect to end up
doing a round of profiling in the future and have to poke some holes in
things too. But until then... safe is just easier and, well, safer!
In the end it'll be lots of detail work, and I really hope it all works
out.
I probably got more than one thing wrong there, and there's layers of
complexity I glossed over, but that's the rough idea ^^
* If we need to fix this then we're going to need some kind of signal from
the DRM core that this is happening and it's not normal user-triggered GEM
closing, and it's going to be interesting... it also means we need some kind
of mechanism to transfer responsibility over those mappings to all in-flight
jobs themselves, because normally userspace is strictly responsible for all
mappings in an explicit sync VM bind style world, and now we're adding a
special case where we freeze the state of the VM until all in-flight jobs
are done when the File goes away instead of eagerly freeing things. That's a
very weird departure from how the whole thing normally works, so if we
really want that I'm going to have to think of how to do it reasonably. It
might be easier once we implement the full VM map range tracking which will
likely flip the VM<->GEM object relationship around a bit.
See my other reply for you how can untangle this potentially on the vm_id
subthread. I think in general rust Drop is awesome, but there's cases
where we'll have to be more explicit at unwinding things. At least in the
linux kernel the general consensus is that too much refcounting is a
maintenance disaster because you end up dropping the final reference in
all kinds of really bad places (interrupt handlers, while holding the
wrong locks, holding them too long).
Yeah, that's one I've run into, it's why I have a few places with subtle
code to avoid that and more things will need cleaning up too...
This is something that I expect will become Rust compile-time magic in
the future though! There's already been talk about being able to prove
that things are only ever called from the right execution context. Then
if you have a Drop of a reference whose underlying type is not safe to
Drop from IRQ context in IRQ context, the compiler complains. It's not
clear exactly what this will look like yet, but it's an active topic of
discussion. And that's another nice thing about Rust: we can actually
help drive the evolution of the language, much faster than with C!
Somewhat related I also expect that
rust drivers will need to have quite a few manual drop calls to
artificially limit lifetime, because sometimes when you drop things the
wrong way round (e.g. drop the refcount while still having the mutex guard
live) results in deadlocks.
Yup! ^^
drivers/gpu/drm/asahi$ grep -r core::mem::drop | wc -l
17
Rust is magic but it won't magically solve all of our problems!
(I don't think all of those 17 are necessary, some are just places where
I wanted to be explicit/safer or bound critical sections more, but a few
are definitely there to avoid deadlocks.)
Simple Arc<Mutex<T>> usage isn't a problem though, you can't drop the
refcount without dropping the guard first there (lifetimes prevent it).
The tricky bit is when you have to call into *another* agent that owns a
reference to the same object, after having done something while holding
the mutex. Then you really need to drop the guard first.
There's going to be lots of fun here :-)
It's already been fun and I don't expect it to get any less fun ^^
~~ Lina