[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20241011065734.471f3774@foz.lan>
Date: Fri, 11 Oct 2024 06:57:34 +0200
From: Mauro Carvalho Chehab <mchehab+huawei@...nel.org>
To: Chenyuan Yang <chenyuan0y@...il.com>
Cc: mchehab@...nel.org, linux-media@...r.kernel.org,
linux-kernel@...r.kernel.org, syzkaller@...glegroups.com, Zijie Zhao
<zzjas98@...il.com>
Subject: Re: [Linux Kernel Bug] memory leak in dvb_dmxdev_add_pid
Em Thu, 10 Oct 2024 20:06:29 -0500
Chenyuan Yang <chenyuan0y@...il.com> escreveu:
> Dear Linux Developers for DVB,
>
> I am writing to inquire if there have been any updates regarding the
> memory leak issue. The issue remains reproducible on the latest stable
> Linux version (6.12-rc2, commit
> 8cf0b93919e13d1e8d4466eb4080a4c4d9d66d7b).
The DVB demux has internally a state machine to filter MPEG-TS streams. It
needs to allocate data and buffers when setting such filters. So, the ioctls
described at https://www.kernel.org/doc/html/v4.10/media/uapi/dvb/dmx_fcalls.html
work together.
Having a report that calling DMX functions on a fuzz testing result on
memory keeping allocated sounds a normal behavior to me, as such filters
are meant to be persistent.
I need to double-check, but I'm almost sure this is persistent even after
device close(). So, de-allocation should happen when the device driver is
removed or when a new set of filters is set. In the last case, the old
buffers will be freed and a new set of buffers will be allocated.
>
> Thank you for your attention to this matter.
>
> Best,
> Chenyuan
>
> On Sat, Mar 2, 2024 at 3:12 PM Chenyuan Yang <chenyuan0y@...il.com> wrote:
> >
> > Dear Linux Developers for DVB,
> >
> > We encountered "memory leak in dvb_dmxdev_add_pid" when testing the
> > DVB driver with Syzkaller and our generated specifications.
> >
> > The C reproducer and the config for the kernel are attached.
> >
> > The memory leak originates from the allocated dmxdev_feed structure,
> > as referenced in the code at
> > [https://elixir.bootlin.com/linux/latest/source/drivers/media/dvb-core/dmxdev.c#L881].
> > This structure fails to be freed upon entering the code branch found
> > at [https://elixir.bootlin.com/linux/latest/source/drivers/media/dvb-core/dmxdev.c#L891].
> >
> > ```
> > ioctl$KGPT_DMX_START(r0, 0x6f29, 0x0)
> > BUG: memory leak
> > unreferenced object 0xffff88802e9ae7e0 (size 32):
> > comm "syz-executor.0", pid 27777, jiffies 4295115050 (age 15.550s)
> > hex dump (first 32 bytes):
> > 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
> > 08 c0 6a 05 00 c9 ff ff 08 c0 6a 05 00 c9 ff ff ..j.......j.....
> > backtrace:
> > [<ffffffff8169126f>] kmemleak_alloc_recursive
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/./include/linux/kmemleak.h:42
> > [inline]
> > [<ffffffff8169126f>] slab_post_alloc_hook
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/mm/slab.h:766
> > [inline]
> > [<ffffffff8169126f>] slab_alloc_node
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/mm/slub.c:3478
> > [inline]
> > [<ffffffff8169126f>] __kmem_cache_alloc_node+0x2ff/0x3e0
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/mm/slub.c:3517
> > [<ffffffff815d9da9>] kmalloc_trace+0x29/0x90
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/mm/slab_common.c:1098
> > [<ffffffff83db2e09>] kmalloc
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/./include/linux/slab.h:600
> > [inline]
> > [<ffffffff83db2e09>] kzalloc
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/./include/linux/slab.h:721
> > [inline]
> > [<ffffffff83db2e09>] dvb_dmxdev_add_pid+0xa9/0x160
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/drivers/media/dvb-core/dmxdev.c:881
> > [<ffffffff83db48de>] dvb_dmxdev_pes_filter_set
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/drivers/media/dvb-core/dmxdev.c:956
> > [inline]
> > [<ffffffff83db48de>] dvb_demux_do_ioctl+0x67e/0xa80
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/drivers/media/dvb-core/dmxdev.c:1076
> > [<ffffffff83db1252>] dvb_usercopy+0x82/0x220
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/drivers/media/dvb-core/dvbdev.c:986
> > [<ffffffff83db1b51>] dvb_demux_ioctl+0x31/0x40
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/drivers/media/dvb-core/dmxdev.c:1185
> > [<ffffffff8171ca88>] vfs_ioctl
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/fs/ioctl.c:51
> > [inline]
> > [<ffffffff8171ca88>] __do_sys_ioctl
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/fs/ioctl.c:871
> > [inline]
> > [<ffffffff8171ca88>] __se_sys_ioctl
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/fs/ioctl.c:857
> > [inline]
> > [<ffffffff8171ca88>] __x64_sys_ioctl+0x108/0x150
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/fs/ioctl.c:857
> > [<ffffffff8540b150>] do_syscall_x64
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/arch/x86/entry/common.c:51
> > [inline]
> > [<ffffffff8540b150>] do_syscall_64+0x40/0x110
> > scratch/zijie-data/LLM-Kernel/spec-eval/shared_linux_builds/syzbot-leak-more_631373bc9e824969/arch/x86/entry/common.c:82
> > [<ffffffff8560008b>] entry_SYSCALL_64_after_hwframe+0x63/0x6b
> > ```
> >
> > If you have any questions or require more information, please feel
> > free to contact us.
> >
> > Reported-by: Chenyuan Yang <chenyuan0y@...il.com>
> >
> > Best,
> > Chenyuan
Thanks,
Mauro
Powered by blists - more mailing lists