Mantis - git
Viewing Issue Advanced Details
ID: Category: Severity: Reproducibility: Date Submitted: Last Update:
5298 regular use major always 2017-02-19 18:02 2017-02-20 10:21
Reporter: danny Platform:  
Assigned To: dam OS:  
Priority: normal OS Version:  
Status: assigned Product Version:  
Product Build: Resolution: open  
Projection: none      
ETA: none Fixed in Version:  
Summary: 0005298: git fast-export producing corrupt output with UTF8 locale on Solaris 11
Description: On Solaris 11, using git 2.3.1 or 2.4.0 from OpenCSW, when commiter or author contain multibyte characters, git fast-export writes extra garbage after the author/committer line:

$ locale
LANG=en_US.UTF8
LC_CTYPE="en_US.UTF8"
[...]
LC_ALL=

$ git fast-export --all
blob
mark :1
data 6
Hello

reset refs/heads/master
commit refs/heads/master
mark :2
author Füü Bär <foo@example.com> 1487506001 +0100
co
committer Füu Bär <foo@example.com> 1487506001 +0100


data 6
Hello
M 100644 :1 foo.txt

reset refs/heads/master
from :2



With LC_CTYPE=C the problem disappears (this correct output can be used with git fast-import to create a test case repository):

$ LC_CTYPE=C git fast-export --all
blob
mark :1
data 6
Hello

reset refs/heads/master
commit refs/heads/master
mark :2
author Füü Bär <foo@example.com> 1487506001 +0100
committer Füu Bär <foo@example.com> 1487506001 +0100
data 6
Hello
M 100644 :1 foo.txt

reset refs/heads/master
from :2
Steps To Reproduce:
Additional Information: The problem might be that git is not compiled in c99 compliant mode. On Solaris 11, this apparently causes printf() to interpret the length field when formatting strings as a number of characters instead of a number of bytes, which results in additional garbage output when the string contains multibyte UTF8 characters.

This following C program illustrates the difference in behaviour:

$ cat printf_test.c
#include <stdio.h>
#include <strings.h>
#include <locale.h>

int main() {
    char test[] = { 102, -61, -74, -61, -74, 98, -61, -92, 114, 64, 64, 64, 0 };
    setlocale(LC_CTYPE, "");
    printf("%.9s\n", test);
    return 0;
}

$ cc -o printf_test printf_test.c
$ ./printf_test
fööbär@@@

$ cc -o printf_test printf_test.c -xc99
$ ./printf_test
fööbär

Or with gcc:

$ gcc -o printf_test printf_test.c
$ ./printf_test
fööbär@@@

$ gcc -o printf_test printf_test.c /usr/lib/32/values-xpg6.o
$ ./printf_test
fööbär
Relationships
Attached Files:
Issue History
Date Modified Username Field Change
2017-02-19 18:02 danny New Issue
2017-02-20 10:21 dam Status new => assigned
2017-02-20 10:21 dam Assigned To => dam

There are no notes attached to this issue.