RE: [PATCH] iommu/iova: kmemleak when disable SRIOV.

["Deucher, Alexander via iommu" <iommu@lists.linux-foundation.org>]

[Public]

> -----Original Message-----
> From: Zhou, Peng Ju <PengJu.Zhou@amd.com>
> Sent: Tuesday, August 3, 2021 1:51 AM
> To: Chris Wilson <chris@chris-wilson.co.uk>; Robin Murphy
> <robin.murphy@arm.com>; iommu@lists.linux-foundation.org
> Cc: Deucher, Alexander <Alexander.Deucher@amd.com>; Wang, Yin
> <Yin.Wang@amd.com>; will@kernel.org; Chang, HaiJun
> <HaiJun.Chang@amd.com>; Deng, Emily <Emily.Deng@amd.com>
> Subject: RE: [PATCH] iommu/iova: kmemleak when disable SRIOV.
> 
> [AMD Official Use Only]
> 
> Hi Chris
> I hit kmemleak with your following patch, Can you help to fix it?
> 
> According to the info in this thread, it seems the patch doesn't merge into
> iommu mainline branch, but I can get your patch from my kernel: 5.11.0

If this patch is not upstream, it probably ended up in our tree via drm-tip or drm-misc last time we synced up.  If that is the case and the patch is not upstream, you can just revert the patch from our tree.

Alex

> 
> 
> commit 48a64dd561a53fb809e3f2210faf5dd442cfc56d
> Author: Chris Wilson <chris@chris-wilson.co.uk>
> Date:   Sat Jan 16 11:10:35 2021 +0000
> 
>     iommu/iova: Use bottom-up allocation for DMA32
> 
>     Since DMA32 allocations are currently allocated top-down from the 4G
>     boundary, this causes fragmentation and reduces the maximise allocation
>     size. On some systems, the dma address space may be extremely
>     constrained by PCIe windows, requiring a stronger anti-fragmentation
>     strategy.
> 
>     Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/2929
>     Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
> 
> 
> ----------------------------------------------------------------------
> BW
> Pengju Zhou
> 
> 
> 
> 
> 
> > -----Original Message-----
> > From: Robin Murphy <robin.murphy@arm.com>
> > Sent: Tuesday, July 27, 2021 10:23 PM
> > To: Zhou, Peng Ju <PengJu.Zhou@amd.com>; iommu@lists.linux-
> > foundation.org
> > Cc: Deucher, Alexander <Alexander.Deucher@amd.com>; Wang, Yin
> > <Yin.Wang@amd.com>; will@kernel.org
> > Subject: Re: [PATCH] iommu/iova: kmemleak when disable SRIOV.
> >
> > On 2021-07-27 15:05, Zhou, Peng Ju wrote:
> > > [AMD Official Use Only]
> > >
> > > Hi Robin
> > > The patch which add "head" :
> > >
> > > commit 48a64dd561a53fb809e3f2210faf5dd442cfc56d
> > > Author: Chris Wilson <chris@chris-wilson.co.uk>
> > > Date:   Sat Jan 16 11:10:35 2021 +0000
> > >
> > >      iommu/iova: Use bottom-up allocation for DMA32
> > >
> > >      Since DMA32 allocations are currently allocated top-down from the 4G
> > >      boundary, this causes fragmentation and reduces the maximise
> allocation
> > >      size. On some systems, the dma address space may be extremely
> > >      constrained by PCIe windows, requiring a stronger anti-fragmentation
> > >      strategy.
> > >
> > >      Closes:
> >
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgitla
> b.f
> > reedesktop.org%2Fdrm%2Fintel%2F-
> >
> %2Fissues%2F2929&amp;data=04%7C01%7CPengJu.Zhou%40amd.com%7C4
> 7f
> >
> c4308f6044a379ed908d9510a19b1%7C3dd8961fe4884e608e11a82d994e183d
> >
> %7C0%7C0%7C637629927137121754%7CUnknown%7CTWFpbGZsb3d8eyJWIj
> o
> >
> iMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C20
> 00
> >
> &amp;sdata=iO5%2FKSW8KV1UZtwGU3oiZpYqiR4eBNcSpF3%2Ft6uSDpY%3D
> &
> > amp;reserved=0
> > >      Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
> >
> > ...which is not in mainline. I've never even seen it posted for review.
> > In fact two search engines can't seem to find any trace of that SHA or patch
> > subject on the internet at all.
> >
> > By all means tell Chris that his patch, wherever you got it from, is broken,
> but
> > once again there's nothing the upstream maintainers/reviewers can do
> about
> > code which isn't upstream.
> >
> > Thanks,
> > Robin.
> >
> > > ----------------------------------------------------------------------
> > > BW
> > > Pengju Zhou
> > >
> > >
> > >
> > >> -----Original Message-----
> > >> From: Robin Murphy <robin.murphy@arm.com>
> > >> Sent: Tuesday, July 27, 2021 4:52 PM
> > >> To: Zhou, Peng Ju <PengJu.Zhou@amd.com>; iommu@lists.linux-
> > >> foundation.org
> > >> Cc: Deucher, Alexander <Alexander.Deucher@amd.com>; Wang, Yin
> > >> <Yin.Wang@amd.com>; will@kernel.org
> > >> Subject: Re: [PATCH] iommu/iova: kmemleak when disable SRIOV.
> > >>
> > >> On 2021-07-27 05:46, Zhou, Peng Ju wrote:
> > >>> [AMD Official Use Only]
> > >>>
> > >>> Hi Robin
> > >>> 1. it is not a good manner to free a statically allocated object(in
> > >>> this case, it
> > >> is iovad->head) dynamically even though the free only occurred when
> > >> shut down the OS in most cases.
> > >>> 2. the kmemleak occurred when disable SRIOV(remove a PCIE device),
> I
> > >>> post the log in the following, in the log, the line:" kmemleak:
> > >>> Found object by alias at 0xffff9221ae647050" means the OS frees a
> > >>> not existing object(iovad->head) which added to RB_TREE in the
> > >>> function init_iova_domain
> > >>
> > >> Sure, it was apparent enough what the bug was; my point is that the
> > >> bug does not exist in mainline. This is the current mainline
> > >> definition of struct
> > >> iova_domain:
> > >>
> > >>
> > >> /* holds all the iova translations for a domain */ struct iova_domain {
> > >> 	spinlock_t	iova_rbtree_lock; /* Lock to protect update of rbtree
> > >> */
> > >> 	struct rb_root	rbroot;		/* iova domain rbtree root */
> > >> 	struct rb_node	*cached_node;	/* Save last alloced node */
> > >> 	struct rb_node	*cached32_node; /* Save last 32-bit alloced node */
> > >> 	unsigned long	granule;	/* pfn granularity for this domain */
> > >> 	unsigned long	start_pfn;	/* Lower limit for this domain */
> > >> 	unsigned long	dma_32bit_pfn;
> > >> 	unsigned long	max32_alloc_size; /* Size of last failed allocation */
> > >> 	struct iova_fq __percpu *fq;	/* Flush Queue */
> > >>
> > >> 	atomic64_t	fq_flush_start_cnt;	/* Number of TLB flushes
> > >> that
> > >> 						   have been started */
> > >>
> > >> 	atomic64_t	fq_flush_finish_cnt;	/* Number of TLB flushes
> > >> that
> > >> 						   have been finished */
> > >>
> > >> 	struct iova	anchor;		/* rbtree lookup anchor */
> > >> 	struct iova_rcache rcaches[IOVA_RANGE_CACHE_MAX_SIZE];	/*
> > >> IOVA range caches */
> > >>
> > >> 	iova_flush_cb	flush_cb;	/* Call-Back function to flush IOMMU
> > >> 					   TLBs */
> > >>
> > >> 	iova_entry_dtor entry_dtor;	/* IOMMU driver specific destructor
> > >> for
> > >> 					   iova entry */
> > >>
> > >> 	struct timer_list fq_timer;		/* Timer to regularily empty
> > >> the
> > >> 						   flush-queues */
> > >> 	atomic_t fq_timer_on;			/* 1 when timer is active, 0
> > >> 						   when not */
> > >> 	struct hlist_node	cpuhp_dead;
> > >> };
> > >>
> > >>
> > >> Please point to where "head" exists either way ;)
> > >>
> > >> The mainline code already has a special case to avoid trying to free
> > >> the statically-allocated "anchor" node. Whoever modified your kernel
> > >> has apparently failed to preserve the equivalent behaviour.
> > >>
> > >> Robin.
> > >>
> > >>> The patch I sent out may don't meet IOMMU requirement because I
> have
> > >> no knowledge of the background of IOMMU, but this patch can fix this
> > >> kmemleak.
> > >>>
> > >>> The kmemleak log on my server:
> > >>> 316613 Jul 19 02:14:20 z-u18 kernel: [  108.967526] pci 0000:83:02.0:
> > >>> Removing from iommu group 108
> > >>> 316614 Jul 19 02:14:20 z-u18 kernel: [  108.969032] kmemleak: Found
> > >> object by alias at 0xffff9221ae647050
> > >>> 316615 Jul 19 02:14:20 z-u18 kernel: [  108.969037] CPU: 30 PID:
> > >>> 2768
> > >> Comm: modprobe Tainted: G           OE     5.11.0 #       12
> > >>> 316616 Jul 19 02:14:20 z-u18 kernel: [  108.969042] Hardware name:
> > >> Supermicro ...
> > >>> 316617 Jul 19 02:14:20 z-u18 kernel: [  108.969045] Call Trace:
> > >>> 316618 Jul 19 02:14:20 z-u18 kernel: [  108.969049]
> > >>> dump_stack+0x6d/0x88
> > >>> 316619 Jul 19 02:14:20 z-u18 kernel: [  108.969061]
> > >>> lookup_object+0x5f/0x70
> > >>> 316620 Jul 19 02:14:20 z-u18 kernel: [  108.969070]
> > >>> find_and_remove_object+0x29/0x80
> > >>> 316621 Jul 19 02:14:20 z-u18 kernel: [  108.969077]
> > >>> delete_object_full+0xc/0x20
> > >>> 316622 Jul 19 02:14:20 z-u18 kernel: [  108.969083]
> > >>> kmem_cache_free+0x22b/0x410
> > >>> 316623 Jul 19 02:14:20 z-u18 kernel: [  108.969097]
> > >>> free_iova_mem+0x22/0x60
> > >>> 316624 Jul 19 02:14:20 z-u18 kernel: [  108.969103]
> > >>> put_iova_domain+0x1b5/0x1e0
> > >>> 316625 Jul 19 02:14:20 z-u18 kernel: [  108.969108]
> > >>> iommu_put_dma_cookie+0x44/0xc0
> > >>> 316626 Jul 19 02:14:20 z-u18 kernel: [  108.969118]
> > >>> domain_exit+0xba/0xc0
> > >>> 316627 Jul 19 02:14:20 z-u18 kernel: [  108.969123]
> > >>> iommu_group_release+0x51/0x90
> > >>> 316628 Jul 19 02:14:20 z-u18 kernel: [  108.969129]
> > >>> kobject_put+0xa7/0x210
> > >>> 316629 Jul 19 02:14:20 z-u18 kernel: [  108.969140]
> > >>> iommu_release_device+0x41/0x80
> > >>> 316630 Jul 19 02:14:20 z-u18 kernel: [  108.969147]
> > >>> iommu_bus_notifier+0xa0/0xc0
> > >>> 316631 Jul 19 02:14:20 z-u18 kernel: [  108.969153]
> > >>> blocking_notifier_call_chain+0x63/0x90
> > >>> 316632 Jul 19 02:14:20 z-u18 kernel: [  108.969162]
> > >>> device_del+0x29c/0x3f0
> > >>> 316633 Jul 19 02:14:20 z-u18 kernel: [  108.969167]
> > >>> pci_remove_bus_device+0x77/0x100
> > >>> 316634 Jul 19 02:14:20 z-u18 kernel: [  108.969178]
> > >>> pci_iov_remove_virtfn+0xbc/0x110
> > >>> 316635 Jul 19 02:14:20 z-u18 kernel: [  108.969187]
> > >>> sriov_disable+0x2f/0xe0 ......
> > >>> 316651 Jul 19 02:14:20 z-u18 kernel: [  108.969629] RIP:
> > >> 0033:0x7f6d8ec45047
> > >>> 316652 Jul 19 02:14:20 z-u18 kernel: [  108.969634] Code: 73 01 c3
> > >>> 48 8b
> > >> 0d 41 8e 2c 00 f7 d8 64 89 01 48 83 c8 ff c3 66        2e 0f 1f 84 00 00 00 00
> > >> 00 0f 1f 44 00 00 b8 b0 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d
> > >> 11 8e 2c 00 f       7 d8 64 89 01 48
> > >>> 316653 Jul 19 02:14:20 z-u18 kernel: [  108.969638] RSP:
> > >> 002b:00007ffc18dafc48 EFLAGS: 00000206 ORIG_RAX: 00000000000000b
> > >> 0
> > >>> 316654 Jul 19 02:14:20 z-u18 kernel: [  108.969645] RAX:
> > >>> ffffffffffffffda RBX: 000055817f00adc0 RCX: 00007f6d8ec45047
> > >>> 316655 Jul 19 02:14:20 z-u18 kernel: [  108.969648] RDX:
> > >>> 0000000000000000 RSI: 0000000000000800 RDI: 000055817f00ae28
> > >>> 316656 Jul 19 02:14:20 z-u18 kernel: [  108.969651] RBP:
> > >>> 000055817f00adc0 R08: 00007ffc18daebf1 R09: 0000000000000000
> > >>> 316657 Jul 19 02:14:20 z-u18 kernel: [  108.969654] R10:
> > >>> 00007f6d8ecc1c40 R11: 0000000000000206 R12: 000055817f00ae28
> > >>> 316658 Jul 19 02:14:20 z-u18 kernel: [  108.969656] R13:
> > >>> 0000000000000001 R14: 000055817f00ae28 R15: 00007ffc18db1030
> > >>> 316659 Jul 19 02:14:20 z-u18 kernel: [  108.969661] kmemleak: Object
> > >> 0xffff9221ae647000 (size 2048):
> > >>> 316660 Jul 19 02:14:20 z-u18 kernel: [  108.969665] kmemleak:   comm
> > >> "gpu_init_thread", pid 2687, jiffies 4294911321
> > >>> 316661 Jul 19 02:14:20 z-u18 kernel: [  108.969669] kmemleak:
> > >> min_count = 1
> > >>> 316662 Jul 19 02:14:20 z-u18 kernel: [  108.969671] kmemleak:   count =
> 0
> > >>> 316663 Jul 19 02:14:20 z-u18 kernel: [  108.969672] kmemleak:   flags =
> > >> 0x1
> > >>> 316664 Jul 19 02:14:20 z-u18 kernel: [  108.969674] kmemleak:
> checksum
> > >> = 0
> > >>> 316665 Jul 19 02:14:20 z-u18 kernel: [  108.969675] kmemleak:
> backtrace:
> > >>> 316666 Jul 19 02:14:20 z-u18 kernel: [  108.969677]
> > >> cookie_alloc+0x1f/0x60
> > >>> 316667 Jul 19 02:14:20 z-u18 kernel: [  108.969682]
> > >> iommu_get_dma_cookie+0x17/0x30
> > >>> 316668 Jul 19 02:14:20 z-u18 kernel: [  108.969685]
> > >> intel_iommu_domain_alloc+0xa7/0xe0
> > >>> 316669 Jul 19 02:14:20 z-u18 kernel: [  108.969689]
> > >> iommu_group_alloc_default_domain+0x4c/0x160
> > >>> 316670 Jul 19 02:14:20 z-u18 kernel: [  108.969695]
> > >> iommu_probe_device+0xff/0x130
> > >>> 316671 Jul 19 02:14:20 z-u18 kernel: [  108.969702]
> > >> iommu_bus_notifier+0x7c/0xc0
> > >>> 316672 Jul 19 02:14:20 z-u18 kernel: [  108.969708]
> > >> blocking_notifier_call_chain+0x63/0x90
> > >>> 316673 Jul 19 02:14:20 z-u18 kernel: [  108.969713]
> > >> device_add+0x3f9/0x860
> > >>> 316674 Jul 19 02:14:20 z-u18 kernel: [  108.969717]
> > >> pci_device_add+0x2c3/0x6a0
> > >>> 316675 Jul 19 02:14:20 z-u18 kernel: [  108.969723]
> > >> pci_iov_add_virtfn+0x1b1/0x300
> > >>> 316676 Jul 19 02:14:20 z-u18 kernel: [  108.969729]
> > >> sriov_enable+0x254/0x410
> > >>>
> > >>>
> > >>> --------------------------------------------------------------------
> > >>> --
> > >>> BW
> > >>> Pengju Zhou
> > >>>
> > >>>
> > >>>
> > >>>
> > >>>> -----Original Message-----
> > >>>> From: Robin Murphy <robin.murphy@arm.com>
> > >>>> Sent: Thursday, July 22, 2021 10:59 PM
> > >>>> To: Zhou, Peng Ju <PengJu.Zhou@amd.com>; iommu@lists.linux-
> > >>>> foundation.org
> > >>>> Cc: Deucher, Alexander <Alexander.Deucher@amd.com>; Wang, Yin
> > >>>> <Yin.Wang@amd.com>; will@kernel.org
> > >>>> Subject: Re: [PATCH] iommu/iova: kmemleak when disable SRIOV.
> > >>>>
> > >>>> On 2021-07-22 09:16, Peng Ju Zhou via iommu wrote:
> > >>>>> the object iova->head allocated statically when enable SRIOV but
> > >>>>> freed dynamically when disable SRIOV which causing kmemleak.
> > >>>>> changing the allocation from statically to dynamically.
> > >>>>
> > >>>> Thanks for the glimpse into the kind of weird and wonderful things
> > >>>> people are doing to the IOVA allocator out-of-tree (the "holes"
> > >>>> list sounds like an idea I also thought about a long time ago), but
> > >>>> judging by the context this patch is clearly of no use to mainline
> > >>>> ;)
> > >>>>
> > >>>> Robin.
> > >>>>
> > >>>>> Signed-off-by: Peng Ju Zhou <PengJu.Zhou@amd.com>
> > >>>>> ---
> > >>>>>     drivers/iommu/iova.c | 15 ++++++++-------
> > >>>>>     include/linux/iova.h |  4 ++--
> > >>>>>     2 files changed, 10 insertions(+), 9 deletions(-)
> > >>>>>
> > >>>>> diff --git a/drivers/iommu/iova.c b/drivers/iommu/iova.c index
> > >>>>> 2371524796d3..505881d8d97f 100644
> > >>>>> --- a/drivers/iommu/iova.c
> > >>>>> +++ b/drivers/iommu/iova.c
> > >>>>> @@ -26,6 +26,8 @@ static void free_iova_rcaches(struct
> iova_domain
> > >>>> *iovad);
> > >>>>>     static void fq_destroy_all_entries(struct iova_domain *iovad);
> > >>>>>     static void fq_flush_timeout(struct timer_list *t);
> > >>>>>     static void free_global_cached_iovas(struct iova_domain
> > >>>>> *iovad);
> > >>>>> +static inline struct iova *alloc_and_init_iova(unsigned long pfn_lo,
> > >>>>> +					       unsigned long pfn_hi);
> > >>>>>
> > >>>>>     void
> > >>>>>     init_iova_domain(struct iova_domain *iovad, unsigned long
> > >>>>> granule, @@ -47,17 +49,16 @@ init_iova_domain(struct
> iova_domain
> > >>>>> *iovad, unsigned long granule,
> > >>>>>
> > >>>>>     	INIT_LIST_HEAD(&iovad->holes);
> > >>>>>
> > >>>>> -	iovad->head.pfn_lo = 0;
> > >>>>> -	iovad->head.pfn_hi = start_pfn;
> > >>>>> -	rb_link_node(&iovad->head.node, NULL, &iovad-
> >rbroot.rb_node);
> > >>>>> -	rb_insert_color(&iovad->head.node, &iovad->rbroot);
> > >>>>> -	list_add(&iovad->head.hole, &iovad->holes);
> > >>>>> +	iovad->head = alloc_and_init_iova(0, start_pfn);
> > >>>>> +	rb_link_node(&iovad->head->node, NULL, &iovad-
> >rbroot.rb_node);
> > >>>>> +	rb_insert_color(&iovad->head->node, &iovad->rbroot);
> > >>>>> +	list_add(&iovad->head->hole, &iovad->holes);
> > >>>>>
> > >>>>>     	iovad->tail.pfn_lo = IOVA_ANCHOR;
> > >>>>>     	iovad->tail.pfn_hi = IOVA_ANCHOR;
> > >>>>>     	rb_link_node(&iovad->tail.node,
> > >>>>> -		     &iovad->head.node,
> > >>>>> -		     &iovad->head.node.rb_right);
> > >>>>> +		     &iovad->head->node,
> > >>>>> +		     &iovad->head->node.rb_right);
> > >>>>>     	rb_insert_color(&iovad->tail.node, &iovad->rbroot);
> > >>>>>
> > >>>>>     	init_iova_rcaches(iovad);
> > >>>>> diff --git a/include/linux/iova.h b/include/linux/iova.h index
> > >>>>> 076eb6cfc613..553905ef41fe 100644
> > >>>>> --- a/include/linux/iova.h
> > >>>>> +++ b/include/linux/iova.h
> > >>>>> @@ -81,7 +81,7 @@ struct iova_domain {
> > >>>>>     						   have been finished
> > */
> > >>>>>
> > >>>>>     	struct list_head holes;
> > >>>>> -	struct iova	head, tail;		/* rbtree lookup
> anchors */
> > >>>>> +	struct iova	*head, tail;		/* rbtree lookup
> anchors */
> > >>>>>     	struct iova_rcache
> rcaches[IOVA_RANGE_CACHE_MAX_SIZE];
> > 	/*
> > >>>> IOVA range caches */
> > >>>>>
> > >>>>>     	iova_flush_cb	flush_cb;	/* Call-Back function to flush
> > IOMMU
> > >>>>> @@ -252,7 +252,7 @@ static inline void
> > >>>>> free_cpu_cached_iovas(unsigned int cpu,
> > >>>>>
> > >>>>>     static inline unsigned long iovad_start_pfn(struct iova_domain
> *iovad)
> > >>>>>     {
> > >>>>> -	return iovad->head.pfn_hi;
> > >>>>> +	return iovad->head->pfn_hi;
> > >>>>>     }
> > >>>>>
> > >>>>>     #endif
> > >>>>>
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu