[Bug locale/31143] New: Let iconv_open return EMFILE if RLIMIT_NOFILE is hit

[Bug locale/31143] New: Let iconv_open return EMFILE if RLIMIT_NOFILE is hit

[jengelh at inai dot de]

https://sourceware.org/bugzilla/show_bug.cgi?id=31143

            Bug ID: 31143
           Summary: Let iconv_open return EMFILE if RLIMIT_NOFILE is hit
           Product: glibc
           Version: 2.38
            Status: UNCONFIRMED
          Severity: enhancement
          Priority: P2
         Component: locale
          Assignee: unassigned at sourceware dot org
          Reporter: jengelh at inai dot de
  Target Milestone: ---

Consider a program that opens a lot of file descriptors, by chance so happens
to stay just barely under the limit, and at some point uses iconv_open.
Pathalogical testcase:

#include <iconv.h>
#include <sys/resource.h>
#include <stdio.h>
int main()
{
        struct rlimit rl = {3,3};
        if (setrlimit(RLIMIT_NOFILE, &rl) != 0) 
                perror("setrlimit");

        iconv_t cd = iconv_open("UTF-8", "windows-1254//IGNORE");
        if (cd == (iconv_t)-1)
                perror("iconv_open");
}

== Observed behavior

On a typical linux-glibc system, this will report:

       iconv_open: Invalid argument

== Expected

       iconv_open: Too many files open

glibc iconv is implemented with shared plugins (/usr/lib/gconv), thus the
implementation is susceptible to fd limits. Though EINVAL is not incorrect per
se (no module loaded - no charset available), it could help the admin to see an
EMFILE errno.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug locale/31143] Let iconv_open return EMFILE if RLIMIT_NOFILE is hit

[fweimer at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=31143

Florian Weimer <fweimer at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Depends on|                            |22040
                 CC|                            |fweimer at redhat dot com

--- Comment #1 from Florian Weimer <fweimer at redhat dot com> ---
This depends on accurate tracking of errno values from system calls during
dlopen (and, ugh, dlsym).


Referenced Bugs:

https://sourceware.org/bugzilla/show_bug.cgi?id=22040
[Bug 22040] dlopen: Provide more information on the root cause of a failure
-- 
You are receiving this mail because:
You are on the CC list for the bug.