On Sat, Dec 01, 2018 at 12:23:51AM +0000, Gong S. wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA256
> 
> >Can you tell what version of the kernel you are using?
> 4.19.0-trunk-amd64. I guess that I am not getting any semblance of support 
> with an experimental kernel.
> However, when I upgrade from 4.19.0-rc7 to 4.19.0-trunk, the problem is still 
> there, so it does not look like a kernel problem.

That certainly doesn't mean that it's not a kernel bug; it just means
that it's a kernel bug that wasn't fixed between 4.19-rc7 and 4.19.

The inline_data feature isn't enabled by default because we're not
100% confidence that it's fully rock solid.  We are regularly running
regression testing on that configuration, and while there are some
test failures with inline_data enabled that aren't there with the
default options, none of them have caused processes to get stuck:

ext4/4k: 444 tests, 2 failures, 42 skipped, 4442 seconds
  Failures: ext4/034 generic/388
ext4/encrypt: 512 tests, 1 failures, 123 skipped, 2638 seconds
  Failures: ext4/034
ext4/dioread_nolock: 443 tests, 2 failures, 42 skipped, 4338 seconds
  Failures: ext4/034 generic/388

vs

ext4/adv: 448 tests, 4 failures, 48 skipped, 4149 seconds
  Failures: ext4/034 generic/399 generic/477 generic/519

So basically, at the moment I can't really recommend this feature for
general use --- although this particular failure which you've reported
is a new one for me.

> >How/when did you enable this feature?
> Just me poking around options in the man page. I assume that I can save some 
> space with small files.
> >It looks like you were trying to upgrade some packages when you ran into 
> >this issue.
> It happened to random processes. It happened to "mandb", "perl" and 
> "chromium" most frequently. Sometimes it happens to very basic processes like 
> "ls" or "rm" (happened once when I try to remove Chromium's cache folder).
> Also, if a process is stuck, the processes will be stuck again if I invoke 
> that program again.

Is this true even if you reboot?   What if you force an fsck check on the file 
system?

                                           - Ted

Reply via email to