git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: Probable bug in file run-command.c function clear_child_for_cleanup

From
Jeff King <peff@peff.net>
Date
Sep 10, 2012, 13:44 UTC
Message-ID
<20120910134406.GA7496@sigill.intra.peff.net>
In-Reply-To
<504CAB66.1050003@optimisefitness.com>
On Sun, Sep 09, 2012 at 03:44:54PM +0100, David Gould wrote:
Show 20 quoted lines
> static void clear_child_for_cleanup(pid_t pid)
> {
> 	struct child_to_clean **last, *p;
> 
> 	last = &children_to_clean;
> 	for (p = children_to_clean; p; p = p->next) {
> 		if (p->pid == pid) {
> 			*last = p->next;
> 			free(p);
> 			return;
> 		}
> 	}
> }
> 
> It appears that last is intended to point to the next field that's
> being updated, but fails to "follow" the p pointer along the chain.
> The result is that children_to_clean will end up pointing to the
> entry after the deleted one, and all the entries before it will be
> lost. It'll only be fine in the case that the pid is that of the
> first entry in the chain.
Yes, it's a bug. We should update "last" on each iteration.
Show 11 quoted lines
> You want something like:
> 
> for (... {
> 	if (... {
> 		...
> 	}
> 	last = &p->next;
> }
> 
> or (probably clearer, but I haven't read your coding style guide, if
> there is one, and some people don't like this approach)
Yes, that's the correct fix. Care to submit a patch?
> for (p = children_to_clean; p; last = &p->next, p = p->next) {
> 	...
That is OK, too, but I think I prefer the first one.
-Peff
Previous: David GouldNext: Erik Faye-Lund
Message 2 of 8 in “Probable bug in file run-command.c function clear_child_for_cleanup”
  1. David GouldSep 9, 2012
  2. Jeff KingSep 10, 2012
  3. Erik Faye-LundSep 10, 2012
  4. Jeff KingSep 10, 2012
  5. Junio C HamanoSep 10, 2012
  6. Jeff KingSep 10, 2012
  7. Junio C HamanoSep 10, 2012
  8. David GouldSep 11, 2012

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.