https://jira.whamcloud.com This file is an XML representation of an issue en-us 9.4.14 940014 05-12-2023 https://jira.whamcloud.com/browse/LU-15591 Lustre <p>In e2fsprogs <tt>ext2fs_open2()</tt> there is a missing <tt>goto cleanup</tt> after <tt>retval = EXT2_ET_SB_CSUM_INVALID</tt>. And probably after <tt>retval = EXT2_ET_UNKNOWN_CSUM;</tt>.</p> <div class="code panel" style="border-width: 1px;"><div class="codeContent panelContent"> <pre class="code-java"> <span class="code-keyword">if</span> (!(fs-&gt;flags &amp; EXT2_FLAG_IGNORE_CSUM_ERRORS)) { retval = 0; <span class="code-keyword">if</span> (!ext2fs_verify_csum_type(fs, fs-&gt;<span class="code-keyword">super</span>)) retval = EXT2_ET_UNKNOWN_CSUM; <span class="code-keyword">if</span> (!ext2fs_superblock_csum_verify(fs, fs-&gt;<span class="code-keyword">super</span>)) { <span class="code-keyword">if</span> (csum_retries++ &lt; 3) <span class="code-keyword">goto</span> retry; retval = EXT2_ET_SB_CSUM_INVALID; } } </pre> </div></div> <p>Also I find it strange that we check the super block magic after verifying the checksum.</p> LU-15591

ext2fs_open2() missing goto

Bug Minor Closed Not a Bug WC Triage John Hammond Wed, 23 Feb 2022 22:21:56 +0000 Fri, 25 Feb 2022 00:36:21 +0000 Fri, 25 Feb 2022 00:36:17 +0000 0 2 <p>I was looking at the history of this code. There used to be an "<tt>if (retval) goto cleanup;</tt>" here, but it was moved:</p> <div class="preformatted panel" style="border-width: 1px;"><div class="preformattedContent panelContent"> <pre>commit f92c600c09bbda482c6c4a37602db4f0db864a9e Author: Darrick J. Wong &lt;darrick.wong@oracle.com&gt; AuthorDate: Mon Sep 8 16:11:49 2014 -0700 Commit: Theodore Ts'o &lt;tytso@mit.edu&gt; CommitDate: Thu Sep 11 12:40:54 2014 -0400 libext2fs: report bad magic over bad sb checksum We don't want ext2fs_open2() to report bad sb checksum on something that's not even an ext* superblock. This apparently happens pretty easily if we try to open an XFS filesystem. Thus, make it so that a bad magic number code always trumps the sb checksum error code. Signed-off-by: Darrick J. Wong &lt;darrick.wong@oracle.com&gt; Signed-off-by: Theodore Ts'o &lt;tytso@mit.edu&gt; diff --git a/lib/ext2fs/openfs.c b/lib/ext2fs/openfs.c index 00320f34..1d6f147c 100644 --- a/lib/ext2fs/openfs.c +++ b/lib/ext2fs/openfs.c @@ -221,8 +221,6 @@ errcode_t ext2fs_open2(const char *name, const char *io_opti ons, retval = EXT2_ET_UNKNOWN_CSUM; if (!ext2fs_superblock_csum_verify(fs, fs-&gt;super)) retval = EXT2_ET_SB_CSUM_INVALID; - if (retval) - goto cleanup; } #ifdef WORDS_BIGENDIAN @@ -235,10 +233,11 @@ errcode_t ext2fs_open2(const char *name, const char *io_options, } #endif - if (fs-&gt;super-&gt;s_magic != EXT2_SUPER_MAGIC) { + if (fs-&gt;super-&gt;s_magic != EXT2_SUPER_MAGIC) retval = EXT2_ET_BAD_MAGIC; + if (retval) goto cleanup; - } + if (fs-&gt;super-&gt;s_rev_level &gt; EXT2_LIB_CURRENT_REV) { retval = EXT2_ET_REV_TOO_HIGH; goto cleanup; </pre> </div></div> <p>It isn't clear why the whole checksum validation block wasn't moved below the superblock magic number check. It may be because the checksum needs to be computed on the un-swabbed/little-endian superblock, which happens immediately after the checksum validation?</p> <p>Is this actually a functional problem, or just an anomaly that you found while looking into this code? Looking at the code it appears it will correctly handle a non-zero checksum <tt>retval</tt> (i.e. it won't be overwritten by 0), just not immediately. I'm just wondering if a fix is needed for this issue for proper <tt>O_DIRECT</tt> usage, or if I should tag e2fsprogs-1.46.2.wc5 with only the <a href="https://jira.whamcloud.com/browse/LU-15590" title="e2fsprogs direct IO raw_read_blk() issues" class="issue-link" data-issue-key="LU-15590"><del>LU-15590</del></a> fix?</p> <p>Maybe it's more confusing that a bug. Looking at the traces, it tried to read a good checksum three times then it returned EXT2_ET_BAD_MAGIC.</p> Related LU-15590 Development Rank 1|i02j9j: Rank (Obsolete) 9223372036854775807 Severity