Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1238132 > unrolled thread

Re: [PATCH] fuse: break infinite loop in fuse_fill_write_pages()

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH] fuse: break infinite loop in fuse_fill_write_pages() Konstantin Khlebnikov <koct9i@gmail.com> - 2015-10-02 13:30 +0200
    Re: [fuse-devel] [PATCH] fuse: break infinite loop in  fuse_fill_write_pages() Maxim Patlasov <mpatlasov@parallels.com> - 2015-10-02 21:30 +0200
      Re: [fuse-devel] [PATCH] fuse: break infinite loop in  fuse_fill_write_pages() Andrew Morton <akpm@linux-foundation.org> - 2015-10-03 00:10 +0200
        Re: [fuse-devel] [PATCH] fuse: break infinite loop in fuse_fill_write_pages() Konstantin Khlebnikov <koct9i@gmail.com> - 2015-10-03 04:00 +0200
          Re: [fuse-devel] [PATCH] fuse: break infinite loop in  fuse_fill_write_pages() Maxim Patlasov <mpatlasov@parallels.com> - 2015-10-05 20:40 +0200
            [PATCH v2] fuse: break infinite loop in fuse_fill_write_pages() Roman Gushchin <klamm@yandex-team.ru> - 2015-10-12 15:40 +0200

#1238132 — Re: [PATCH] fuse: break infinite loop in fuse_fill_write_pages()

Bump. Add more peopple in CC.

On Mon, Sep 21, 2015 at 1:02 PM, Roman Gushchin <klamm@yandex-team.ru> wrote:
> I got a report about unkillable task eating CPU. Thge further
> investigation shows, that the problem is in the fuse_fill_write_pages()
> function. If iov's first segment has zero length, we get an infinite
> loop, because we never reach iov_iter_advance() call.
>
> Fix this by calling iov_iter_advance() before repeating an attempt to
> copy data from userspace.
>
> A similar problem is described in 124d3b7041f ("fix writev regression:
> pan hanging unkillable and un-straceable").
>
> Signed-off-by: Roman Gushchin <klamm@yandex-team.ru>
> Cc: Miklos Szeredi <miklos@szeredi.hu>
> ---
>  fs/fuse/file.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/fs/fuse/file.c b/fs/fuse/file.c
> index f523f2f..195476a 100644
> --- a/fs/fuse/file.c
> +++ b/fs/fuse/file.c
> @@ -1049,6 +1049,7 @@ static ssize_t fuse_fill_write_pages(struct fuse_req *req,
>                 tmp = iov_iter_copy_from_user_atomic(page, ii, offset, bytes);
>                 flush_dcache_page(page);
>
> +               iov_iter_advance(ii, tmp);
>                 if (!tmp) {
>                         unlock_page(page);
>                         page_cache_release(page);
> @@ -1061,7 +1062,6 @@ static ssize_t fuse_fill_write_pages(struct fuse_req *req,
>                 req->page_descs[req->num_pages].length = tmp;
>                 req->num_pages++;
>
> -               iov_iter_advance(ii, tmp);
>                 count += tmp;
>                 pos += tmp;
>                 offset += tmp;
> --
> 2.4.3
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [next] | [standalone]

#1238552 — Re: [fuse-devel] [PATCH] fuse: break infinite loop in fuse_fill_write_pages()

On 10/02/2015 04:21 AM, Konstantin Khlebnikov wrote:
> Bump. Add more peopple in CC.
>
> On Mon, Sep 21, 2015 at 1:02 PM, Roman Gushchin <klamm@yandex-team.ru> wrote:
>> I got a report about unkillable task eating CPU. Thge further
>> investigation shows, that the problem is in the fuse_fill_write_pages()
>> function. If iov's first segment has zero length, we get an infinite
>> loop, because we never reach iov_iter_advance() call.

iov_iter_copy_from_user_atomic() eventually calls iterate_iovec(). The 
latter silently consumes zero-length iov. So I don't think "iov's first 
segment has zero length" can cause infinite loop.

Thanks,
Maxim

>>
>> Fix this by calling iov_iter_advance() before repeating an attempt to
>> copy data from userspace.
>>
>> A similar problem is described in 124d3b7041f ("fix writev regression:
>> pan hanging unkillable and un-straceable").
>>
>> Signed-off-by: Roman Gushchin <klamm@yandex-team.ru>
>> Cc: Miklos Szeredi <miklos@szeredi.hu>
>> ---
>>   fs/fuse/file.c | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/fs/fuse/file.c b/fs/fuse/file.c
>> index f523f2f..195476a 100644
>> --- a/fs/fuse/file.c
>> +++ b/fs/fuse/file.c
>> @@ -1049,6 +1049,7 @@ static ssize_t fuse_fill_write_pages(struct fuse_req *req,
>>                  tmp = iov_iter_copy_from_user_atomic(page, ii, offset, bytes);
>>                  flush_dcache_page(page);
>>
>> +               iov_iter_advance(ii, tmp);
>>                  if (!tmp) {
>>                          unlock_page(page);
>>                          page_cache_release(page);
>> @@ -1061,7 +1062,6 @@ static ssize_t fuse_fill_write_pages(struct fuse_req *req,
>>                  req->page_descs[req->num_pages].length = tmp;
>>                  req->num_pages++;
>>
>> -               iov_iter_advance(ii, tmp);
>>                  count += tmp;
>>                  pos += tmp;
>>                  offset += tmp;
>> --
>> 2.4.3
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>> Please read the FAQ at  http://www.tux.org/lkml/
> ------------------------------------------------------------------------------
> _______________________________________________
> fuse-devel mailing list
> fuse-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/fuse-devel

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1238651 — Re: [fuse-devel] [PATCH] fuse: break infinite loop in fuse_fill_write_pages()

On Fri, 2 Oct 2015 12:27:45 -0700 Maxim Patlasov <mpatlasov@parallels.com> wrote:

> On 10/02/2015 04:21 AM, Konstantin Khlebnikov wrote:
> > Bump. Add more peopple in CC.
> >
> > On Mon, Sep 21, 2015 at 1:02 PM, Roman Gushchin <klamm@yandex-team.ru> wrote:
> >> I got a report about unkillable task eating CPU. Thge further
> >> investigation shows, that the problem is in the fuse_fill_write_pages()
> >> function. If iov's first segment has zero length, we get an infinite
> >> loop, because we never reach iov_iter_advance() call.
> 
> iov_iter_copy_from_user_atomic() eventually calls iterate_iovec(). The 
> latter silently consumes zero-length iov. So I don't think "iov's first 
> segment has zero length" can cause infinite loop.

I'm suspecting it got stuck because local variable `bytes' is zero, so
the code does `goto again' repeatedly.

Or maybe not.   A more complete description of the bug would help. 
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1238717 — Re: [fuse-devel] [PATCH] fuse: break infinite loop in fuse_fill_write_pages()

On Sat, Oct 3, 2015 at 1:04 AM, Andrew Morton <akpm@linux-foundation.org> wrote:
> On Fri, 2 Oct 2015 12:27:45 -0700 Maxim Patlasov <mpatlasov@parallels.com> wrote:
>
>> On 10/02/2015 04:21 AM, Konstantin Khlebnikov wrote:
>> > Bump. Add more peopple in CC.
>> >
>> > On Mon, Sep 21, 2015 at 1:02 PM, Roman Gushchin <klamm@yandex-team.ru> wrote:
>> >> I got a report about unkillable task eating CPU. Thge further
>> >> investigation shows, that the problem is in the fuse_fill_write_pages()
>> >> function. If iov's first segment has zero length, we get an infinite
>> >> loop, because we never reach iov_iter_advance() call.
>>
>> iov_iter_copy_from_user_atomic() eventually calls iterate_iovec(). The
>> latter silently consumes zero-length iov. So I don't think "iov's first
>> segment has zero length" can cause infinite loop.
>
> I'm suspecting it got stuck because local variable `bytes' is zero, so
> the code does `goto again' repeatedly.
>
> Or maybe not.   A more complete description of the bug would help.

I suspect here is the same scenario like in 124d3b7041f:
Zero-length segmend is followed by segment with invalid address:
iov_iter_fault_in_readable() checks only first segment (zero-length)
iov_iter_copy_from_user_atomic() skips it, fails at second and
returns zero -> goto again without skipping zero-length segment.

Patch calls iov_iter_advance() before goto again: we'll skip zero-length
segment at second iteraction and iov_iter_fault_in_readable() will detect
invalid address.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1239832 — Re: [fuse-devel] [PATCH] fuse: break infinite loop in fuse_fill_write_pages()

On 10/02/2015 06:58 PM, Konstantin Khlebnikov wrote:
> On Sat, Oct 3, 2015 at 1:04 AM, Andrew Morton <akpm@linux-foundation.org> wrote:
>> On Fri, 2 Oct 2015 12:27:45 -0700 Maxim Patlasov <mpatlasov@parallels.com> wrote:
>>
>>> On 10/02/2015 04:21 AM, Konstantin Khlebnikov wrote:
>>>> Bump. Add more peopple in CC.
>>>>
>>>> On Mon, Sep 21, 2015 at 1:02 PM, Roman Gushchin <klamm@yandex-team.ru> wrote:
>>>>> I got a report about unkillable task eating CPU. Thge further
>>>>> investigation shows, that the problem is in the fuse_fill_write_pages()
>>>>> function. If iov's first segment has zero length, we get an infinite
>>>>> loop, because we never reach iov_iter_advance() call.
>>> iov_iter_copy_from_user_atomic() eventually calls iterate_iovec(). The
>>> latter silently consumes zero-length iov. So I don't think "iov's first
>>> segment has zero length" can cause infinite loop.
>> I'm suspecting it got stuck because local variable `bytes' is zero, so
>> the code does `goto again' repeatedly.
>>
>> Or maybe not.   A more complete description of the bug would help.
> I suspect here is the same scenario like in 124d3b7041f:
> Zero-length segmend is followed by segment with invalid address:
> iov_iter_fault_in_readable() checks only first segment (zero-length)
> iov_iter_copy_from_user_atomic() skips it, fails at second and
> returns zero -> goto again without skipping zero-length segment.
>
> Patch calls iov_iter_advance() before goto again: we'll skip zero-length
> segment at second iteraction and iov_iter_fault_in_readable() will detect
> invalid address.

Makes sense to me. The patch looks fine.

Thanks,
Maxim
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [next] | [standalone]

#1244691 — [PATCH v2] fuse: break infinite loop in fuse_fill_write_pages()

I got a report about unkillable task eating CPU. Further
investigation shows, that the problem is in the fuse_fill_write_pages()
function. If iov's first segment has zero length, we get an infinite
loop, because we never reach iov_iter_advance() call.

Fix this by calling iov_iter_advance() before repeating an attempt to
copy data from userspace.

A similar problem is described in 124d3b7041f ("fix writev regression:
pan hanging unkillable and un-straceable"). If zero-length segmend
is followed by segment with invalid address,
iov_iter_fault_in_readable() checks only first segment (zero-length),
iov_iter_copy_from_user_atomic() skips it, fails at second and
returns zero -> goto again without skipping zero-length segment.

Patch calls iov_iter_advance() before goto again: we'll skip zero-length
segment at second iteraction and iov_iter_fault_in_readable() will detect
invalid address.

Special thanks to Konstantin Khlebnikov, who helped a lot with the commit
description.

Signed-off-by: Roman Gushchin <klamm@yandex-team.ru>
Cc: Miklos Szeredi <miklos@szeredi.hu>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Maxim Patlasov <mpatlasov@parallels.com>
Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Signed-off-by: Roman Gushchin <klamm@yandex-team.ru>
---
 fs/fuse/file.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/fuse/file.c b/fs/fuse/file.c
index f523f2f..195476a 100644
--- a/fs/fuse/file.c
+++ b/fs/fuse/file.c
@@ -1049,6 +1049,7 @@ static ssize_t fuse_fill_write_pages(struct fuse_req *req,
 		tmp = iov_iter_copy_from_user_atomic(page, ii, offset, bytes);
 		flush_dcache_page(page);
 
+		iov_iter_advance(ii, tmp);
 		if (!tmp) {
 			unlock_page(page);
 			page_cache_release(page);
@@ -1061,7 +1062,6 @@ static ssize_t fuse_fill_write_pages(struct fuse_req *req,
 		req->page_descs[req->num_pages].length = tmp;
 		req->num_pages++;
 
-		iov_iter_advance(ii, tmp);
 		count += tmp;
 		pos += tmp;
 		offset += tmp;
-- 
2.4.3

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web