OpenZFS 9403 - assertion failed in arc_buf_destroy()

Assertion failed in arc_buf_destroy() when concurrently reading
block with checksum error.

Porting notes:
* The ability to zinject decompression errors has been added, but
  this only works at the zio_decompress() level, where we have all
  of the info we need to match against the user's zinject options.
* The decompress_fault test has been added to test the new zinject
  functionality
* We attempted to set zio_decompress_fail_fraction to (1 << 18) in
  ztest for further test coverage. Although this did uncover a few
  low priority issues, this unfortuantely also causes ztest to
  ASSERT in many locations where the code is working correctly since
  it is designed to fail on IO errors. Developers can manually set
  this variable with the '-o' option to find and debug issues.

Authored by: Matt Ahrens <[email protected]>
Reviewed by: George Wilson <[email protected]>
Reviewed by: Paul Dagnelie <[email protected]>
Reviewed by: Pavel Zakharov <[email protected]>
Reviewed by: Brian Behlendorf <[email protected]>
Approved by: Matt Ahrens <[email protected]>
Ported-by: Tom Caputi <[email protected]>

OpenZFS-issue: https://illumos.org/issues/9403
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/fa98e487a9
Closes #7822 

1 files changed, 6 insertions, 4 deletions

diff --git a/cmd/zinject/zinject.c b/cmd/zinject/zinject.c
index d6298f8f4..ee15ff8a4 100644
--- a/cmd/zinject/zinject.c
+++ b/cmd/zinject/zinject.c
@@ -116,9 +116,9 @@
  * specified.
  *
  * The '-e' option takes a string describing the errno to simulate.  This must
- * be one of 'io', 'checksum', or 'decrypt'.  In most cases this will result
- * in the same behavior, but RAID-Z will produce a different set of ereports
- * for this situation.
+ * be one of 'io', 'checksum', 'decompress', or 'decrypt'.  In most cases this
+ * will result in the same behavior, but RAID-Z will produce a different set of
+ * ereports for this situation.
  *
  * The '-a', '-u', and '-m' flags toggle internal flush behavior.  If '-a' is
  * specified, then the ARC cache is flushed appropriately.  If '-u' is
@@ -300,7 +300,7 @@ usage(void)
 	    "\n"
 	    "\t\t-q\tQuiet mode.  Only print out the handler number added.\n"
 	    "\t\t-e\tInject a specific error.  Must be one of 'io',\n"
-	    "\t\t\t'checksum', or decrypt.  Default is 'io'.\n"
+	    "\t\t\t'checksum', 'decompress', or decrypt.  Default is 'io'.\n"
 	    "\t\t-l\tInject error at a particular block level. Default is "
 	    "0.\n"
 	    "\t\t-m\tAutomatically remount underlying filesystem.\n"
@@ -774,6 +774,8 @@ main(int argc, char **argv)
 				error = EIO;
 			} else if (strcasecmp(optarg, "checksum") == 0) {
 				error = ECKSUM;
+			} else if (strcasecmp(optarg, "decompress") == 0) {
+				error = EINVAL;
 			} else if (strcasecmp(optarg, "decrypt") == 0) {
 				error = EACCES;
 			} else if (strcasecmp(optarg, "nxio") == 0) {