linux-ext4 - Re: [GIT PULL] e4defrag inline data segfault fix

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20260117044904.GD15522@frogsfrogsfrogs>
Date: Fri, 16 Jan 2026 20:49:04 -0800
From: "Darrick J. Wong" <djwong@...nel.org>
To: Theodore Tso <tytso@....edu>
Cc: Daniel Tang <danielzgtg.opensource@...il.com>,
	linux-ext4@...r.kernel.org
Subject: Re: [GIT PULL] e4defrag inline data segfault fix

On Fri, Jan 16, 2026 at 05:47:47PM -1000, Theodore Tso wrote:
> On Fri, Jan 16, 2026 at 06:35:59PM -0800, Darrick J. Wong wrote:
> > > -	/* Has no blocks */
> > > -	if (buf->st_blocks == 0) {
> > > +	/* Has 0 or 1 blocks, no point to defragment */
> > > +	if (buf->st_blocks <= buf->st_blksize / 512) {
> > 
> > ...because can't you call FS_IOC_GETFLAGS and look for
> > EXT4_INLINE_DATA_FL?
> 
> I could have checked for EXT4_INLINE_DATA_FL, but we need to call
> stat(2) to check for the st_blocks == 0 case, and while it is harmless
> to defrag a file with a single data block, it's also pointless and a
> waste of system calls.  So it's best that we skip defragging the file
> in these cases:
> 
>   A) A zero-length file with st_blocks == 0
>   B) A file with a single data block (st_blocks == st_blksize / 512)
>   C) A file with inline data (st_blocks == 1)
> 
> ... and we can do that only by checking the values returned by
> stat(2).
> 
> Yes, (B) and (C) relies on Linux's behavior, since Posix is silent on
> the semantics of st_blocks, but e4defrag works only by using a
> Linux-specific ioctl, and using FS_IOC_GETFLAGS would also be
> Linux-only.

Fair enough.

Reviewed-by: "Darrick J. Wong" <djwong@...nel.org>

--D


>     	       	       	       		- Ted
>