Re: [syzbot] [block?] general protection fault in bio_first_folio

From: Matthew Wilcox
Date: Sat Dec 16 2023 - 00:53:30 EST


On Fri, Dec 15, 2023 at 09:20:22PM -0800, syzbot wrote:
> Hello,
>
> syzbot found the following issue on:

Good bot.

> commit 1b151e2435fc3a9b10c8946c6aebe9f3e1938c55
> Author: Matthew Wilcox (Oracle) <willy@xxxxxxxxxxxxx>
> Date: Mon Aug 14 14:41:00 2023 +0000
>
> block: Remove special-casing of compound pages
>
> bisection log: https://syzkaller.appspot.com/x/bisect.txt?x=152f8a8ee80000
> final oops: https://syzkaller.appspot.com/x/report.txt?x=172f8a8ee80000
> console output: https://syzkaller.appspot.com/x/log.txt?x=132f8a8ee80000
>
> general protection fault, probably for non-canonical address 0xdffffc0000000001: 0000 [#1] PREEMPT SMP KASAN
> KASAN: null-ptr-deref in range [0x0000000000000008-0x000000000000000f]

I _think_ what's happened here is that __bio_release_pages() was called
with a zero-length bio.

> CPU: 0 PID: 5059 Comm: syz-executor696 Not tainted 6.7.0-rc5-next-20231212-syzkaller #0
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 11/10/2023
> RIP: 0010:_compound_head include/linux/page-flags.h:247 [inline]
> RIP: 0010:bio_first_folio+0xcb/0x5c0 include/linux/bio.h:289

... bio_first_folio() doesn't handle it well, whereas
bio_for_each_segment_all() did.

> If you want syzbot to run the reproducer, reply with:
> #syz test: git://repo/address.git branch-or-commit-hash
> If you attach or paste a git patch, syzbot will apply it before testing.

#syz test git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next abb240f7a2bd

diff --git a/include/linux/bio.h b/include/linux/bio.h
index ec4db73e5f4e..1518f1201ddd 100644
--- a/include/linux/bio.h
+++ b/include/linux/bio.h
@@ -286,6 +286,11 @@ static inline void bio_first_folio(struct folio_iter *fi, struct bio *bio,
{
struct bio_vec *bvec = bio_first_bvec_all(bio) + i;

+ if (i >= bio->bi_vcnt) {
+ fi->folio = NULL;
+ return;
+ }
+
fi->folio = page_folio(bvec->bv_page);
fi->offset = bvec->bv_offset +
PAGE_SIZE * (bvec->bv_page - &fi->folio->page);
@@ -303,10 +308,8 @@ static inline void bio_next_folio(struct folio_iter *fi, struct bio *bio)
fi->offset = 0;
fi->length = min(folio_size(fi->folio), fi->_seg_count);
fi->_next = folio_next(fi->folio);
- } else if (fi->_i + 1 < bio->bi_vcnt) {
- bio_first_folio(fi, bio, fi->_i + 1);
} else {
- fi->folio = NULL;
+ bio_first_folio(fi, bio, fi->_i + 1);
}
}