bpf.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: John Hubbard <jhubbard@nvidia.com>
To: Leon Romanovsky <leon@kernel.org>, Jason Gunthorpe <jgg@ziepe.ca>
Cc: "Christoph Hellwig" <hch@infradead.org>,
	"Andrew Morton" <akpm@linux-foundation.org>,
	"Al Viro" <viro@zeniv.linux.org.uk>,
	"Alex Williamson" <alex.williamson@redhat.com>,
	"Benjamin Herrenschmidt" <benh@kernel.crashing.org>,
	"Björn Töpel" <bjorn.topel@intel.com>,
	"Dan Williams" <dan.j.williams@intel.com>,
	"Daniel Vetter" <daniel@ffwll.ch>,
	"Dave Chinner" <david@fromorbit.com>,
	"David Airlie" <airlied@linux.ie>,
	"David S . Miller" <davem@davemloft.net>,
	"Ira Weiny" <ira.weiny@intel.com>, "Jan Kara" <jack@suse.cz>,
	"Jens Axboe" <axboe@kernel.dk>,
	"Jonathan Corbet" <corbet@lwn.net>,
	"Jérôme Glisse" <jglisse@redhat.com>,
	"Magnus Karlsson" <magnus.karlsson@intel.com>,
	"Mauro Carvalho Chehab" <mchehab@kernel.org>,
	"Michael Ellerman" <mpe@ellerman.id.au>,
	"Michal Hocko" <mhocko@suse.com>,
	"Mike Kravetz" <mike.kravetz@oracle.com>,
	"Paul Mackerras" <paulus@samba.org>,
	"Shuah Khan" <shuah@kernel.org>,
	"Vlastimil Babka" <vbabka@suse.cz>,
	bpf@vger.kernel.org, dri-devel@lists.freedesktop.org,
	kvm@vger.kernel.org, linux-block@vger.kernel.org,
	linux-doc@vger.kernel.org, linux-fsdevel@vger.kernel.org,
	linux-kselftest@vger.kernel.org, linux-media@vger.kernel.org,
	linux-rdma@vger.kernel.org, linuxppc-dev@lists.ozlabs.org,
	netdev@vger.kernel.org, linux-mm@kvack.org,
	LKML <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH v7 07/24] IB/umem: use get_user_pages_fast() to pin DMA pages
Date: Sun, 24 Nov 2019 16:05:16 -0800	[thread overview]
Message-ID: <e8319590-a3f0-5ba4-af4c-65213358a742@nvidia.com> (raw)
In-Reply-To: <20191124100724.GH136476@unreal>

On 11/24/19 2:07 AM, Leon Romanovsky wrote:
> On Thu, Nov 21, 2019 at 10:36:43AM -0400, Jason Gunthorpe wrote:
>> On Thu, Nov 21, 2019 at 12:07:46AM -0800, Christoph Hellwig wrote:
>>> On Wed, Nov 20, 2019 at 11:13:37PM -0800, John Hubbard wrote:
>>>> And get rid of the mmap_sem calls, as part of that. Note
>>>> that get_user_pages_fast() will, if necessary, fall back to
>>>> __gup_longterm_unlocked(), which takes the mmap_sem as needed.
>>>>
>>>> Reviewed-by: Jan Kara <jack@suse.cz>
>>>> Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
>>>> Reviewed-by: Ira Weiny <ira.weiny@intel.com>
>>>> Signed-off-by: John Hubbard <jhubbard@nvidia.com>
>>>
>>> Looks fine,
>>>
>>> Reviewed-by: Christoph Hellwig <hch@lst.de>
>>>
>>> Jason, can you queue this up for 5.5 to reduce this patch stack a bit?
>>
>> Yes, I said I'd do this in an earlier revision. Now that it is clear this
>> won't go through Andrew's tree, applied to rdma for-next
> 
> Jason,
> 
> This patch broke RDMA completely.
> Change from get_user_pages() to get_user_pages_fast() causes to endless
> amount of splats due to combination of the following code:
> 
> 189 struct ib_umem *ib_umem_get(struct ib_udata *udata, unsigned long addr,
> 190                             size_t size, int access)
> ...
> 263         if (!umem->writable)
> 264                 gup_flags |= FOLL_FORCE;
> 265
> 
> and
> 
> 2398 int get_user_pages_fast(unsigned long start, int nr_pages,
> 2399                         unsigned int gup_flags, struct page **pages)
> 2400 {
> 2401         unsigned long addr, len, end;
> 2402         int nr = 0, ret = 0;
> 2403
> 2404         if (WARN_ON_ONCE(gup_flags & ~(FOLL_WRITE | FOLL_LONGTERM)))
> 2405                 return -EINVAL;
> 

Hi Leon,

I looked into this, and I believe that the problem is in gup.c. There appears to
have been an oversight, in commit 817be129e6f2 ("mm: validate get_user_pages_fast
flags"), in filtering out FOLL_FORCE. There is nothing in the _fast() implementation
that requires that we avoid writing to the pages.

So I intend, to post a two-patch series that includes this fix, first, but maybe this
means that it should go in via -mm. I'm not sure what's the best tree.

diff --git a/mm/gup.c b/mm/gup.c
index 8f236a335ae9..745b4036cdfd 100644
--- a/mm/gup.c
+++ b/mm/gup.c
@@ -2401,7 +2401,8 @@ int get_user_pages_fast(unsigned long start, int nr_pages,
 	unsigned long addr, len, end;
 	int nr = 0, ret = 0;
 
-	if (WARN_ON_ONCE(gup_flags & ~(FOLL_WRITE | FOLL_LONGTERM)))
+	if (WARN_ON_ONCE(gup_flags & ~(FOLL_WRITE | FOLL_LONGTERM |
+				       FOLL_FORCE)))
 		return -EINVAL;
 
 	start = untagged_addr(start) & PAGE_MASK;


thanks,
-- 
John Hubbard
NVIDIA


> 
> [   72.623266] ------------[ cut here ]------------
> [   72.624143] WARNING: CPU: 1 PID: 2557 at mm/gup.c:2404 get_user_pages_fast+0x115/0x180
> [   72.625426] Modules linked in: xt_MASQUERADE iptable_nat nf_nat
>    br_netfilter overlay ib_srp scsi_transport_srp rpcrdma rdma_ucm
>    ib_iser libiscsi scsi_transport_iscsi rdma_cm ib_umad iw_cm ib_ipoib
>    ib_cm mlx5_ib ib_uverbs ib_core mlx5_core mlxfw ptp pps_core
> [   72.629024] CPU: 1 PID: 2557 Comm: python3 Not tainted 5.4.0-rc5-J6120-Geeb831ec5b80 #1
> [   72.630435] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1ubuntu1 04/01/2014
> [   72.631973] RIP: 0010:get_user_pages_fast+0x115/0x180
> [   72.632873] Code: 8d 0c 10 85 c0 89 d0 0f 49 c1 eb 97 fa 4c 8d
>        44 24 0c 4c 89 e9 44 89 e2 48 89 df e8 25 dd ff ff fb 8b 44 24 0c
>        e9 75 ff ff ff <0f> 0b b8 ea ff ff ff e9 6d ff ff ff 65 4c 8b 2c
>        25 00 5d 01 00 49
> [   72.635967] RSP: 0018:ffffc90000edf930 EFLAGS: 00010202
> [   72.636896] RAX: 0000000000000000 RBX: 00007f931c392000 RCX: ffff8883f4909000
> [   72.638117] RDX: 0000000000010011 RSI: 0000000000000001 RDI: 00007f931c392000
> [   72.639353] RBP: 00007f931c392000 R08: 0000000000000020 R09: ffff8883f4909000
> [   72.640602] R10: 0000000000000000 R11: ffff8884068d6760 R12: ffff888409e45ba0
> [   72.641858] R13: ffff8884068d6760 R14: 0000160000000000 R15: ffff8883f4909000
> [   72.643115] FS:  00007f9323754700(0000) GS:ffff88842fa80000(0000) knlGS:0000000000000000
> [   72.644586] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [   72.645628] CR2: 00007f931c3ca000 CR3: 00000003f83d6006 CR4: 00000000007606a0
> [   72.646875] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> [   72.648120] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
> [   72.649374] PKRU: 55555554
> [   72.649935] Call Trace:
> [   72.650477]  ib_umem_get+0x298/0x550 [ib_uverbs]
> [   72.651347]  mlx5_ib_db_map_user+0xad/0x130 [mlx5_ib]
> [   72.652279]  mlx5_ib_create_cq+0x1e8/0xaa0 [mlx5_ib]
> [   72.653207]  create_cq+0x1c8/0x2d0 [ib_uverbs]
> [   72.654050]  ib_uverbs_create_cq+0x70/0xa0 [ib_uverbs]
> [   72.654988]  ib_uverbs_handler_UVERBS_METHOD_INVOKE_WRITE+0xc2/0xf0 [ib_uverbs]
> [   72.656331]  ib_uverbs_cmd_verbs.isra.6+0x5be/0xbe0 [ib_uverbs]
> [   72.657398]  ? uverbs_disassociate_api+0xd0/0xd0 [ib_uverbs]
> [   72.658423]  ? kvm_clock_get_cycles+0xd/0x10
> [   72.659229]  ? kmem_cache_alloc+0x176/0x1c0
> [   72.660025]  ? filemap_map_pages+0x18c/0x350
> [   72.660838]  ib_uverbs_ioctl+0xc0/0x120 [ib_uverbs]
> [   72.661756]  do_vfs_ioctl+0xa1/0x610
> [   72.662458]  ksys_ioctl+0x70/0x80
> [   72.663119]	__x64_sys_ioctl+0x16/0x20
> [   72.663850]  do_syscall_64+0x42/0x110
> [   72.664562]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
> [   72.665492] RIP: 0033:0x7f9332958267
> [   72.666185] Code: b3 66 90 48
> 	8b 05 19 3c 2c 00 64 c7 00 26 00
> 	00 00 48 c7 c0 ff ff ff ff c3 66
> 	2e 0f 1f 84 00 00 00 00 00 b8 10
> 	00 00 00 0f 05 <48> 3d 01 f0 ff
> 	ff 73 01 c3 48 8b 0d e9 3b 2c 00
> 	f7 d8 64 89 01 48
> [   72.669347] RSP: 002b:00007f9323751928 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
> [   72.670706] RAX: ffffffffffffffda RBX: 00007f93237519b8 RCX:	00007f9332958267
> [   72.671937] RDX: 00007f93237519a0 RSI: 00000000c0181b01 RDI: 0000000000000008
> [   72.673176] RBP: 00007f9323751980 R08: 0000000000000005 R09: 00007f931c3ffef0
> [   72.674415] R10: 0000000000001000 R11: 0000000000000246 R12: 00007f931c400030
> [   72.675653] R13: 00007f9323751980 R14: 00007f9323751ea8 R15: 00000000000000ff
> [   72.676885] ---[ end trace 033465440517865c ]---
> 
>>
>> Thanks,
>> Jason
>>

  reply	other threads:[~2019-11-25  0:05 UTC|newest]

Thread overview: 55+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-11-21  7:13 [PATCH v7 00/24] mm/gup: track dma-pinned pages: FOLL_PIN John Hubbard
2019-11-21  7:13 ` [PATCH v7 01/24] mm/gup: pass flags arg to __gup_device_* functions John Hubbard
2019-11-21  8:06   ` Christoph Hellwig
2019-11-21  8:25     ` John Hubbard
2019-11-21  7:13 ` [PATCH v7 02/24] mm/gup: factor out duplicate code from four routines John Hubbard
2019-11-21  8:03   ` Christoph Hellwig
2019-11-21  8:29     ` John Hubbard
2019-11-21  9:49       ` Jan Kara
2019-11-21 21:47         ` John Hubbard
2019-11-21  9:54       ` Jan Kara
2019-11-22  2:54         ` John Hubbard
2019-11-22 11:15           ` Jan Kara
2019-11-21  7:13 ` [PATCH v7 03/24] mm/gup: move try_get_compound_head() to top, fix minor issues John Hubbard
2019-11-21  8:07   ` Christoph Hellwig
2019-11-21  7:13 ` [PATCH v7 04/24] mm: Cleanup __put_devmap_managed_page() vs ->page_free() John Hubbard
2019-11-21  8:04   ` Christoph Hellwig
2019-11-21  7:13 ` [PATCH v7 05/24] mm: devmap: refactor 1-based refcounting for ZONE_DEVICE pages John Hubbard
2019-11-21  8:05   ` Christoph Hellwig
2019-11-21  8:54     ` John Hubbard
2019-11-21 16:59       ` Dan Williams
2019-11-21 22:22         ` John Hubbard
2019-11-21  7:13 ` [PATCH v7 06/24] goldish_pipe: rename local pin_user_pages() routine John Hubbard
2019-11-21  8:08   ` Christoph Hellwig
2019-11-21  8:36     ` John Hubbard
2019-11-21  7:13 ` [PATCH v7 07/24] IB/umem: use get_user_pages_fast() to pin DMA pages John Hubbard
2019-11-21  8:07   ` Christoph Hellwig
2019-11-21 14:36     ` Jason Gunthorpe
2019-11-24  6:14       ` John Hubbard
2019-11-24 10:07       ` Leon Romanovsky
2019-11-25  0:05         ` John Hubbard [this message]
2019-11-25  0:53           ` Jason Gunthorpe
2019-11-21  7:13 ` [PATCH v7 08/24] media/v4l2-core: set pages dirty upon releasing DMA buffers John Hubbard
2019-11-21  8:09   ` Christoph Hellwig
2019-11-21  7:13 ` [PATCH v7 09/24] vfio, mm: fix get_user_pages_remote() and FOLL_LONGTERM John Hubbard
2019-11-21  8:10   ` Christoph Hellwig
2019-11-21  8:48     ` John Hubbard
2019-11-21 21:35   ` Alex Williamson
2019-11-21 21:49     ` John Hubbard
2019-11-21  7:13 ` [PATCH v7 10/24] mm/gup: introduce pin_user_pages*() and FOLL_PIN John Hubbard
2019-11-21  7:13 ` [PATCH v7 11/24] goldish_pipe: convert to pin_user_pages() and put_user_page() John Hubbard
2019-11-21  7:13 ` [PATCH v7 12/24] IB/{core,hw,umem}: set FOLL_PIN via pin_user_pages*(), fix up ODP John Hubbard
2019-11-21  7:13 ` [PATCH v7 13/24] mm/process_vm_access: set FOLL_PIN via pin_user_pages_remote() John Hubbard
2019-11-21  7:13 ` [PATCH v7 14/24] drm/via: set FOLL_PIN via pin_user_pages_fast() John Hubbard
2019-11-21  7:13 ` [PATCH v7 15/24] fs/io_uring: set FOLL_PIN via pin_user_pages() John Hubbard
2019-11-21  7:13 ` [PATCH v7 16/24] net/xdp: " John Hubbard
2019-11-21  7:13 ` [PATCH v7 17/24] mm/gup: track FOLL_PIN pages John Hubbard
2019-11-21  9:39   ` Jan Kara
2019-11-30 18:59   ` kbuild test robot
2019-11-21  7:13 ` [PATCH v7 18/24] media/v4l2-core: pin_user_pages (FOLL_PIN) and put_user_page() conversion John Hubbard
2019-11-21  7:13 ` [PATCH v7 19/24] vfio, mm: " John Hubbard
2019-11-21  7:13 ` [PATCH v7 20/24] powerpc: book3s64: convert to pin_user_pages() and put_user_page() John Hubbard
2019-11-21  7:13 ` [PATCH v7 21/24] mm/gup_benchmark: use proper FOLL_WRITE flags instead of hard-coding "1" John Hubbard
2019-11-21  7:13 ` [PATCH v7 22/24] mm/gup_benchmark: support pin_user_pages() and related calls John Hubbard
2019-11-21  7:13 ` [PATCH v7 23/24] selftests/vm: run_vmtests: invoke gup_benchmark with basic FOLL_PIN coverage John Hubbard
2019-11-21  7:13 ` [PATCH v7 24/24] mm, tree-wide: rename put_user_page*() to unpin_user_page*() John Hubbard

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=e8319590-a3f0-5ba4-af4c-65213358a742@nvidia.com \
    --to=jhubbard@nvidia.com \
    --cc=airlied@linux.ie \
    --cc=akpm@linux-foundation.org \
    --cc=alex.williamson@redhat.com \
    --cc=axboe@kernel.dk \
    --cc=benh@kernel.crashing.org \
    --cc=bjorn.topel@intel.com \
    --cc=bpf@vger.kernel.org \
    --cc=corbet@lwn.net \
    --cc=dan.j.williams@intel.com \
    --cc=daniel@ffwll.ch \
    --cc=davem@davemloft.net \
    --cc=david@fromorbit.com \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=hch@infradead.org \
    --cc=ira.weiny@intel.com \
    --cc=jack@suse.cz \
    --cc=jgg@ziepe.ca \
    --cc=jglisse@redhat.com \
    --cc=kvm@vger.kernel.org \
    --cc=leon@kernel.org \
    --cc=linux-block@vger.kernel.org \
    --cc=linux-doc@vger.kernel.org \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-kselftest@vger.kernel.org \
    --cc=linux-media@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=linux-rdma@vger.kernel.org \
    --cc=linuxppc-dev@lists.ozlabs.org \
    --cc=magnus.karlsson@intel.com \
    --cc=mchehab@kernel.org \
    --cc=mhocko@suse.com \
    --cc=mike.kravetz@oracle.com \
    --cc=mpe@ellerman.id.au \
    --cc=netdev@vger.kernel.org \
    --cc=paulus@samba.org \
    --cc=shuah@kernel.org \
    --cc=vbabka@suse.cz \
    --cc=viro@zeniv.linux.org.uk \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).