bug-bash
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [BUG] shell_start_time


From: Chet Ramey
Subject: Re: [BUG] shell_start_time
Date: Mon, 29 Sep 2025 12:13:17 -0400
User-agent: Mozilla Thunderbird

On 9/28/25 8:27 PM, Pourko via Bug reports for the GNU Bourne Again SHell wrote:
execute_cmd.c assumes that the value in time_t shell_start_time is in ticks,

Yes, in that particular case, this is a problem.

as how one would get it from a call to the times() function. But for some
reason, shell.c puts seconds in it, rather than ticks.
It should be seconds, since it backs $SECONDS and other things that the
shell uses with seconds granularity.


Moreover, execute_cmd.c uses times() as a fallbask for !defined 
(HAVE_GETTIMEOFDAY),
yet shell.c uses a call to gettimeofday() to put those seconds in 
shell_start_time.

Bash provides a replacement version of gettimeofday() in lib/sh/gettimeofday.c.

That, however, only fills in the seconds and is useless for command timing.
In execute_cmd.c this bug breaks things in "if (posixly_correct && nullcmd)".

Are you on a system that doesn't provide getrusage()? That's required by
POSIX.

In any case, it should be shell_start_time * get_clk_tck() in that case
to make sure tbefore is in CLK_TCKs and fulfill that peculiar requirement,
since POSIX requires a command argument. It still won't be very granular.

--
``The lyf so short, the craft so long to lerne.'' - Chaucer
                 ``Ars longa, vita brevis'' - Hippocrates
Chet Ramey, UTech, CWRU    chet@case.edu    http://tiswww.cwru.edu/~chet/

Attachment: OpenPGP_signature.asc
Description: OpenPGP digital signature


reply via email to

[Prev in Thread] Current Thread [Next in Thread]