[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: musl compatibility
From: |
Paolo Bonzini |
Subject: |
Re: musl compatibility |
Date: |
Sat, 23 Jun 2012 17:10:54 +0200 |
On Sun, Jun 17, 2012 at 6:15 PM, Bruno Haible <address@hidden> wrote:
> Paul Eggert wrote:
>> On 06/12/2012 04:21 AM, Paolo Bonzini wrote:
>> > perhaps we can follow the suggestion and
>> > replace "if (freadahead (f))" with "if (freading(f) && !feof(f))" in
>> > closein.c.
>>
>> Yes, thanks, I like this idea the best of those suggested so far.
>> Here's a proposed patch to gnulib.
>
> Unfortunately, this idea will cause extra system calls for programs which
> terminate without having read from stdin.
A wrongly-nonzero freadahead is a correctness problem indeed.
However, in the case of close_stdin, is this important for something
that happens rarely, and even then at most once in a run (and
usually the time the lseek will happen anyway, because in general
stdin is read in its entirety)?
Not to mention that lseek(0, 0, SEEK_CUR) is one of the most optimized
system calls in the kernel, possibly beaten only by
gettimeofday/clock_gettime.
Paolo
- musl compatibility, Reuben Thomas, 2012/06/07
- Re: musl compatibility, Pedro Alves, 2012/06/08
- Re: musl compatibility, Reuben Thomas, 2012/06/08
- Re: musl compatibility, Paolo Bonzini, 2012/06/12
- Re: musl compatibility, Paul Eggert, 2012/06/12
- Re: musl compatibility, Bruno Haible, 2012/06/17
- Re: musl compatibility,
Paolo Bonzini <=
- Re: musl compatibility, Paul Eggert, 2012/06/23