2023-12-01 00:01:39

by Brian Norris

[permalink] [raw]
Subject: [PATCH] lib/ext2fs: Validity checks for ext2fs_inode_scan_goto_blockgroup()

We don't validate the 'group' argument, so it's easy to get underflows
or crashes here.

This resolves issues seen in ureadahead, when it uses an old packfile
(with mismatching group indices) with a new filesystem.

Signed-off-by: Brian Norris <[email protected]>
---

lib/ext2fs/inode.c | 3 +++
1 file changed, 3 insertions(+)

diff --git a/lib/ext2fs/inode.c b/lib/ext2fs/inode.c
index 957d5aa9f9d6..96d854b5fb69 100644
--- a/lib/ext2fs/inode.c
+++ b/lib/ext2fs/inode.c
@@ -313,6 +313,9 @@ static errcode_t get_next_blockgroup(ext2_inode_scan scan)
errcode_t ext2fs_inode_scan_goto_blockgroup(ext2_inode_scan scan,
int group)
{
+ if (group <= 0 || group >= scan->fs->group_desc_count)
+ return EXT2_ET_INVALID_ARGUMENT;
+
scan->current_group = group - 1;
scan->groups_left = scan->fs->group_desc_count - group;
scan->bad_block_ptr = 0;
--
2.43.0.rc2.451.g8631bc7472-goog



2023-12-01 16:24:16

by Darrick J. Wong

[permalink] [raw]
Subject: Re: [PATCH] lib/ext2fs: Validity checks for ext2fs_inode_scan_goto_blockgroup()

On Thu, Nov 30, 2023 at 04:01:18PM -0800, Brian Norris wrote:
> We don't validate the 'group' argument, so it's easy to get underflows
> or crashes here.
>
> This resolves issues seen in ureadahead, when it uses an old packfile
> (with mismatching group indices) with a new filesystem.

Say what now? The boot time pre-caching thing Ubuntu used to have?
https://manpages.ubuntu.com/manpages/trusty/man8/ureadahead.8.html

--D

> Signed-off-by: Brian Norris <[email protected]>
> ---
>
> lib/ext2fs/inode.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/lib/ext2fs/inode.c b/lib/ext2fs/inode.c
> index 957d5aa9f9d6..96d854b5fb69 100644
> --- a/lib/ext2fs/inode.c
> +++ b/lib/ext2fs/inode.c
> @@ -313,6 +313,9 @@ static errcode_t get_next_blockgroup(ext2_inode_scan scan)
> errcode_t ext2fs_inode_scan_goto_blockgroup(ext2_inode_scan scan,
> int group)
> {
> + if (group <= 0 || group >= scan->fs->group_desc_count)
> + return EXT2_ET_INVALID_ARGUMENT;
> +
> scan->current_group = group - 1;
> scan->groups_left = scan->fs->group_desc_count - group;
> scan->bad_block_ptr = 0;
> --
> 2.43.0.rc2.451.g8631bc7472-goog
>
>

2023-12-01 17:31:10

by Brian Norris

[permalink] [raw]
Subject: Re: [PATCH] lib/ext2fs: Validity checks for ext2fs_inode_scan_goto_blockgroup()

On Fri, Dec 1, 2023 at 8:24 AM Darrick J. Wong <[email protected]> wrote:
> On Thu, Nov 30, 2023 at 04:01:18PM -0800, Brian Norris wrote:
> > This resolves issues seen in ureadahead, when it uses an old packfile
> > (with mismatching group indices) with a new filesystem.
>
> Say what now? The boot time pre-caching thing Ubuntu used to have?
> https://manpages.ubuntu.com/manpages/trusty/man8/ureadahead.8.html

Sure. ChromeOS still uses it. Steven Rostedt even bothered to do a
talk about it recently:
https://eoss2023.sched.com/event/1LcMw/the-resurrection-of-ureadahead-and-speeding-up-the-boot-process-and-preloading-applications-steven-rostedt-google

Brian

2023-12-02 17:10:51

by Darrick J. Wong

[permalink] [raw]
Subject: Re: [PATCH] lib/ext2fs: Validity checks for ext2fs_inode_scan_goto_blockgroup()

On Fri, Dec 01, 2023 at 09:30:38AM -0800, Brian Norris wrote:
> On Fri, Dec 1, 2023 at 8:24 AM Darrick J. Wong <[email protected]> wrote:
> > On Thu, Nov 30, 2023 at 04:01:18PM -0800, Brian Norris wrote:
> > > This resolves issues seen in ureadahead, when it uses an old packfile
> > > (with mismatching group indices) with a new filesystem.
> >
> > Say what now? The boot time pre-caching thing Ubuntu used to have?
> > https://manpages.ubuntu.com/manpages/trusty/man8/ureadahead.8.html
>
> Sure. ChromeOS still uses it. Steven Rostedt even bothered to do a
> talk about it recently:
> https://eoss2023.sched.com/event/1LcMw/the-resurrection-of-ureadahead-and-speeding-up-the-boot-process-and-preloading-applications-steven-rostedt-google

Wow. I had no idea that ureadahead reads the inode blocks of a mounted
ext* filesystem into the page cache. Welp, it's a good thing those are
part of the static layout.

Anyway, this fix looks correct to me, so I don't see any reason to hold
this up...
Reviewed-by: Darrick J. Wong <[email protected]>

--D

>
> Brian
>