Bug 497267

Summary:

Bug in SHIFT_JIS EILSEQ handling

Product:

[Fedora] Fedora

Reporter:

Behdad Esfahbod <behdad>

Component:

glibc

Assignee:

Jakub Jelinek <jakub>

Status:

CLOSED RAWHIDE

QA Contact:

Fedora Extras Quality Assurance <extras-qa>

Severity:

medium

Docs Contact:

Priority:

low

Version:

rawhide

CC:

jakub

Target Milestone:

---

Target Release:

---

Hardware:

All

OS:

Linux

Whiteboard:

Fixed In Version:

Doc Type:

Bug Fix

Doc Text:

Story Points:

---

Clone Of:

Environment:

Last Closed:

2009-04-27 20:11:21 UTC

Type:

---

Regression:

---

Mount Type:

---

Documentation:

---

CRM:

Verified Versions:

Category:

---

oVirt Team:

---

RHEL 7.3 requirements from Atomic Host:

Cloudforms Team:

---

Target Upstream Version:

Embargoed:

Attachments:

Description	Flags
test case	none

Description Behdad Esfahbod 2009-04-23 05:18:29 UTC

Trying to convert the invalid sequence "\x98\x8c" from SHIFT_JIS to UTF-8, iconv correctly returns -1 and sets errno to EILSEQ, however it advances input pointers, consuming the ill sequence.  It should not.

This is causing gnome-terminal to crash seeing that sequence at the end of input.  I'm fixing the crasher in g-t, but the real bug is in iconv.  g-t bug:
http://bugzilla.gnome.org/show_bug.cgi?id=567064

Attaching test case.

Comment 1 Jakub Jelinek 2009-04-23 10:22:31 UTC

Testcase is missing...

Comment 2 Jakub Jelinek 2009-04-23 10:32:55 UTC

I've tried:
#include <iconv.h>
#include <assert.h>
#include <errno.h>

int
main (void)
{
  iconv_t i1 = iconv_open ("ISO-8859-1//", "UTF-8//");
  iconv_t i2 = iconv_open ("UTF-8//", "SHIFT-JIS//");
  char *inbuf, *outbuf, buf[64];
  size_t inbytesleft, outbytesleft, res;

  inbuf = "\x83\x83\x83";
  inbytesleft = 3;
  outbuf = buf;
  outbytesleft = 64;
  errno = 0;
  res = iconv (i1, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
  assert (res == (size_t) -1 && errno == EILSEQ && inbytesleft == 3 && outbytesleft == 64);

  inbuf = "\xec\x98\x8c";
  inbytesleft = 3;
  outbuf = buf;
  outbytesleft = 64;
  errno = 0;
  res = iconv (i2, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
  assert (res == (size_t) -1 && errno == EILSEQ && inbytesleft == 3 && outbytesleft == 64);

  return 0;
}

but that works as expected.

Comment 3 Behdad Esfahbod 2009-04-23 22:51:47 UTC

Created attachment 341034 [details]
test case

Sorry.  I'm sure I attached in the original report form...

Anyway, seems to work for "\xec\x98\x8c", but not for "x\x98\x8c".

Comment 4 Jakub Jelinek 2009-04-27 20:11:21 UTC

Should be fixed in glibc-2.9.90-22.

Comment 5 Behdad Esfahbod 2009-04-27 20:21:49 UTC

Thanks.