Re: xargs: exec's with too many arguments

[Dmitry V. Levin]

On Tue, Nov 22, 2005 at 11:35:15AM +0000, Tim Waugh wrote:
> On Sat, Nov 19, 2005 at 08:30:50AM +0000, James Youngman wrote:
> 
> > Thanks for discovering and fixing the bug.
> 
> With 4.2.26 I'm finding a test suite failure with the new tests on
> x86_64:

Same here.

> Running 
> /tmp/twaugh/findutils-4.2.26/xargs/testsuite/xargs.posix/arg_max_32bit_linux_bug.exp
>  ...
> FAIL: arg_max_32bit_linux_bug, wrong stdout output: xargs.out 
> /tmp/twaugh/findutils-4.2.26/xargs/testsuite/xargs.posix/arg_max_32bit_linux_bug.xo
>  differ: byte 32764, line 1
> child process exited abnormally
> Running 
> /tmp/twaugh/findutils-4.2.26/xargs/testsuite/xargs.posix/arg_max_64bit_linux_bug.exp
>  ...
> FAIL: arg_max_64bit_linux_bug, wrong stdout output: xargs.out 
> /tmp/twaugh/findutils-4.2.26/xargs/testsuite/xargs.posix/arg_max_64bit_linux_bug.xo
>  differ: byte 32764, line 1
> child process exited abnormally
> 
> The reason is that the expected output contains one line of 16383
> 'y's.  Of course, it is surely meant to be one line of 16382 'y's
> followed by another line with just 'y' -- otherwise the fix hasn't
> worked.
> 
> Same goes for the 32bit case.

Output of these two tests appear to be arch-dependent.

> HOWEVER, I have a 32-bit (x86) Linux installation here which still
> fails arg_max_32bit_linux_bug:

Probably this behavior is kernel-dependent...
Is this behavior kernel-dependent?


-- 
ldv

pgpSXCjSTi_B8.pgp
Description: PGP signature