Bug 433 - mdev -s should not die processing one device node
Summary: mdev -s should not die processing one device node
Status: RESOLVED FIXED
Alias: None
Product: Busybox
Classification: Unclassified
Component: Other (show other bugs)
Version: 1.14.x
Hardware: PC Linux
: P5 minor
Target Milestone: ---
Assignee: unassigned
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2009-07-02 09:22 UTC by Michael Tokarev
Modified: 2009-07-11 22:20 UTC (History)
1 user (show)

See Also:
Host:
Target:
Build:


Attachments
Fix (956 bytes, patch)
2009-07-02 11:16 UTC, Denys Vlasenko
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Michael Tokarev 2009-07-02 09:22:57 UTC
There are still several foo_or_die() calls in make_device().  Like syntax errors in a single line, or failure to run an external command, etc.  Having in mind `mdev -s' mode of operations, each such _die() is dangerous - we're risking leaving system without important device nodes in /dev and hence make it unbootable.
Comment 1 Michael Tokarev 2009-07-02 09:26:15 UTC
See also #309
Comment 2 Denys Vlasenko 2009-07-02 11:16:40 UTC
Created attachment 433 [details]
Fix

This patch makes two errors non-fatal. It does not make mknod() error non-fatal. Do you think it should allow that too?
Comment 3 Denys Vlasenko 2009-07-02 11:18:41 UTC
(In reply to comment #1)
> See also #309

I thought it is fixed. Is it? If not, on what version of bbox do you observe it?
Comment 4 Michael Tokarev 2009-07-02 11:28:19 UTC
(In reply to comment #3)
> (In reply to comment #1)
> > See also #309
> 
> I thought it is fixed. Is it? If not, on what version of bbox do you observe it?

The mention of #390 were just for reference.  It's fixed and is marked as such, but the nature of current #433 is the same as of #309.

(In reply to comment #2)
> Created an attachment (id=433) [details]
> Fix

Yeah, that't what I mean.  Heck, libb contains so many functions... ;)

> This patch makes two errors non-fatal. It does not make mknod() error non-fatal. Do you think it should allow that too?

Strictly speaking yes, but well, the situation when mknod/mkdir returns failure is very rare.  If it's easy enough to do (like, no need to un-wrap something like "mknod_or_die"), let it be fixed too.  If not, let's leave it as it is.