This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: check-abi failure with sys_errlist


On Fri, 6 Apr 2012, Andreas Schwab wrote:

> "Joseph S. Myers" <joseph@codesourcery.com> writes:
> 
> > On Fri, 6 Apr 2012, Andreas Schwab wrote:
> >
> >> 	* sysdeps/gnu/errlist-compat.awk (END): Correct definition of
> >> 	ERR_MAX symbol.
> >> 	* sysdeps/unix/sysv/linux/Versions (GLIBC_2.12): Adjust
> >> 	errlist-compat value.
> >
> > Could you please clarify what the effects of the bug you are fixing, and 
> > of this patch, are?
> 
> ERR_MAX is the highest error number, not the number of entries.

But how is it ERR_MAX used?  How does the ABI depend on ERR_MAX?  What are 
the errlist-compat values meant to mean?  In what circumstances did they 
fail to have the intended meaning?

> > You change one of the errlist-compat entries (but not 
> > others or the SPARC version) in sysdeps/unix/sysv/linux/Versions; how does 
> > one tell whether any other Versions file (whether in libc or ports) also 
> > needs adjusting for this bug fix to avoid it causing any problems (whether 
> > ABI compatibility or otherwise), and what adjustments to make if so?
> 
> None of the other compat entries ever had been inflated.

What is "inflated" here - representing a sys_errlist with room to grow 
beyond the largest current errno value?

Please make it easy for people to understand patches you post, and their 
rationale, by including such explanations with the patch submission (or, 
as appropriate, writing a wiki page explaining the issue and pointing to 
that - a wiki page explaining these issues and what to do when adding a 
new errno value seems like something that would be of use beyond just 
dealing with this patch).

A patch posting with just ChangeLog entry and diffs is very rarely going 
to be appropriate; there will almost always be more explanation needed 
than that to make it easy for other people to see why the patch is correct 
and desirable.  See Carlos's comments in 
<http://sourceware.org/ml/libc-alpha/2012-04/msg00175.html>.

-- 
Joseph S. Myers
joseph@codesourcery.com


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]