Re: [syzbot] BUG: unable to handle kernel paging request in percpu_counter_add_batch

[Sun, Jiebin]

On 1/1/1970 8:00 AM, Andrew Morton wrote:
> On Thu, 20 Oct 2022 19:24:40 -0700 syzbot
> <syzbot+96e659d35b9d6b541152@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
>
> > Hello,
> >
> > syzbot found the following issue on:
> >
> > HEAD commit:    aae703b02f92 Merge tag 'for-6.1-rc1-tag' of git://git.kern..
> > git tree:       upstream
> > console output:
> > https://syzkaller.appspot.com/x/log.txt?x=1686e478880000
> > kernel config:
> > https://syzkaller.appspot.com/x/.config?x=883de36878e6d869
> > dashboard link: https://syzkaller.appspot.com/bug?extid=96e659d35b9d6b541152
> > compiler:       aarch64-linux-gnu-gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for
> Debian) 2.35.2
> > userspace arch: arm
> >
> > Unfortunately, I don't have any reproducer for this issue yet.
> >
> > IMPORTANT: if you fix the issue, please add the following tag to the commit:
> > Reported-by: syzbot+96e659d35b9d6b541152@xxxxxxxxxxxxxxxxxxxxxxxxx
> >
> > Unable to handle kernel paging request at virtual address
> > ffff70000b978a00
> > KASAN: maybe wild-memory-access in range
> > [0xffff80005cbc5000-0xffff80005cbc5007]
> > Mem abort info:
> >    ESR = 0x0000000096000006
> >    EC = 0x25: DABT (current EL), IL = 32 bits
> >    SET = 0, FnV = 0
> >    EA = 0, S1PTW = 0
> >    FSC = 0x06: level 2 translation fault Data abort info:
> >    ISV = 0, ISS = 0x00000006
> >    CM = 0, WnR = 0
> > swapper pgtable: 4k pages, 48-bit VAs, pgdp=0000000045dfb000
> > [ffff70000b978a00] pgd=00000000bfbeb003, p4d=00000000bfbeb003,
> > pud=00000000bfbea003, pmd=0000000000000000 Internal error: Oops:
> > 0000000096000006 [#1] PREEMPT SMP Modules linked in:
> > CPU: 0 PID: 2820 Comm: kworker/0:3 Tainted: G        W          6.1.0-rc1-syzkaller-00025-gaae703b02f92
> #0
> > Hardware name: linux,dummy-virt (DT)
> > Workqueue: events free_ipc
> > pstate: 10000005 (nzcV daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--) pc :
> > percpu_counter_add_batch+0x78/0x2f0 lib/percpu_counter.c:87 lr :
> > percpu_counter_add_batch+0x44/0x2f0 lib/percpu_counter.c:87 sp :
> > ffff8000120979c0
> > x29: ffff8000120979c0 x28: ffff0000151ff800 x27: 1fffe00002a3ff24
> > x26: ffff80005cbc5000 x25: ffff00000eeac7b0 x24: 1fffe00001dd58f6
> > x23: 000000007fffffff x22: ffff80000cc4e800 x21: 0000000000000000
> > x20: 0000000000000000 x19: ffff00000eeac758 x18: ffff00006a9cbbc0
> > x17: 0000000000000000 x16: 0000000000000000 x15: ffff800008d9c830
> > x14: ffff800008d78ecc x13: ffff800008d752d4 x12: ffff600002a3ff01
> > x11: 1fffe00002a3ff00 x10: ffff600002a3ff00 x9 : dfff800000000000
> > x8 : ffff0000151ff800 x7 : 00000000f1f1f1f1 x6 : dfff800000000000
> > x5 : ffff700002412f32 x4 : 1ffff000020ef1ac x3 : 1ffff0000b978a00
> > x2 : dfff800000000000 x1 : 0000000000000003 x0 : ffff80005cbc5000 Call
> > trace:
> >   percpu_counter_add_batch+0x78/0x2f0 lib/percpu_counter.c:87
> > percpu_counter_add_local include/linux/percpu_counter.h:75 [inline]
> > percpu_counter_sub_local include/linux/percpu_counter.h:225 [inline]
> >   freeque+0x20c/0x364 ipc/msg.c:292
> >   free_ipcs+0xa0/0x160 ipc/namespace.c:125
> >   msg_exit_ns+0x38/0x60 ipc/msg.c:1334
> >   free_ipc_ns ipc/namespace.c:138 [inline]
> >   free_ipc+0xd0/0x1c0 ipc/namespace.c:157
> > process_one_work+0x780/0x184c kernel/workqueue.c:2289
> >   worker_thread+0x3cc/0xc40 kernel/workqueue.c:2436
> >   kthread+0x23c/0x2a0 kernel/kthread.c:376
> >   ret_from_fork+0x10/0x20 arch/arm64/kernel/entry.S:860
> Well.  freeque() messes with ns->percpu_msg_hdrs but msg_exit_ns() already did
> percpu_counter_destroy(&ns->percpu_msg_hdrs);
>
> I'm thinking something like this?
>
> --- a/ipc/msg.c~a
> +++ a/ipc/msg.c
> @@ -1329,11 +1329,11 @@ fail_msg_bytes:
>   #ifdef CONFIG_IPC_NS
>   void msg_exit_ns(struct ipc_namespace *ns)  {
> -	percpu_counter_destroy(&ns->percpu_msg_bytes);
> -	percpu_counter_destroy(&ns->percpu_msg_hdrs);
>   	free_ipcs(ns, &msg_ids(ns), freeque);
>   	idr_destroy(&ns->ids[IPC_MSG_IDS].ipcs_idr);
>   	rhashtable_destroy(&ns->ids[IPC_MSG_IDS].key_ht);
> +	percpu_counter_destroy(&ns->percpu_msg_bytes);
> +	percpu_counter_destroy(&ns->percpu_msg_hdrs);
>   }
>   #endif
>   
> _

Hi Andrew,

The quick fix looks good to me. There is protection in 
percpu_counter_destroy if the it has been released in freeque. Thanks.