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

Bug#618033: marked as done (texlive-bin-2009: Missing -D_FILE_OFFSET_BITS=64 leads to files not opening (x86 /32bit))

Your message dated Sun, 4 Aug 2019 21:27:00 +0200
with message-id <49896d3a-f284-da57-7472-a77f9553b1fd@web.de>
and subject line Re: Bug#618033: texlive-bin-2009: Missing -D_FILE_OFFSET_BITS=64 leads to files not opening (x86 /32bit)
has caused the Debian Bug report #618033,
regarding texlive-bin-2009: Missing -D_FILE_OFFSET_BITS=64 leads to files not opening (x86 /32bit)
to be marked as done.

This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.

(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact owner@bugs.debian.org
immediately.)


-- 
618033: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=618033
Debian Bug Tracking System
Contact owner@bugs.debian.org with problems
--- Begin Message --- 
Package: texlive-binaries
Version: 2009-8
Severity: normal

Dear all,

	thank you for all the effort putting a distriubtion together.

After upgrading to the lastly relased stable, I found that pdflatex would
not find (and therefore compile) the files given on the command line,
which were mounted by CIFS.
The error occurs in the kpathsea library (texk/kpathsea/readable.c), where the stat fails with the errno set to EOVERFLOW. It fails because the offset is to large. After I added manually -D_FILE_OFFSET_BITS=64 (following the stat 2 man page), kpathsea/pdflatex finds the file and compiles it happily. I added "a patch", which illustrates my hack to find and circumvent the problem.
Could that be that AC_SYS_LARGEFILE is missing in configure.ac while it is listed in common.ac.orig? 

I am using
Linux 2.6.32-5-686 #1 SMP Wed Jan 12 04:01:41 UTC 2011 i686 GNU/Linux 
and
libc6-i686     2.11.2-10

Happy to answer any further questions.

Sincerely yours
	Pierre Schnizer

--
+---------------------------------------------------------------------+
  Pierre Schnizer  <p.schnizer@gsi.de>

GSI Helmholtzzentrum fuer  Schwerionenforschung GmbH
Planckstrasse 1
D-64291 Darmstadt
www.gsi.de

Gesellschaft mit beschraenkter Haftung
Sitz der Gesellschaft: Darmstadt
Handelsregister: Amtsgericht Darmstadt, HRB 1528

Geschaeftsfuehrung: Professor Dr. Dr. h.c. Horst Stoecker, Dr. Hartmut Eickhoff
Vorsitzende des Aufsichtsrates: Dr. Beatrix Vierkorn-Rudolph
Stellvertreter: Ministerialdirigent Dr. Rolf Bernhardt
+---------------------------------------------------------------------+
--- texlive-bin-2009/debian/rules	2011-03-12 21:23:52.000000000 +0100
+++ texlive-bin-2009-modified/debian/rules	2011-03-13 16:49:37.000000000 +0100
@@ -18,6 +18,15 @@
 ifneq (,$(filter $(DEB_BUILD_ARCH),$(GCC_44_ARCHES)))
     export CC=gcc-4.4
     export CXX=g++-4.4
+else
+    # This is a hack here. I used it to rebuild the binaries on x86/32bit and 
+    # check that the stat() call would then succeed in 
+    # texlive-bin-2009/texk/kpathsea/readable.c
+    # I guess that necessity for the flag should be detected by autconf/automake
+    # I am not familiar enough with the GNU buildchain to try that
+    # Pierre
+    export CC=gcc -D_FILE_OFFSET_BITS=64
+    export CXX=g++ -D_FILE_OFFSET_BITS=64
 endif
 
 ifneq (,$(filter $(DEB_BUILD_ARCH),$(RELAX_ARCHES)))
--- texlive-bin-2009/texk/kpathsea/readable.c	2009-03-16 16:13:07.000000000 +0100
+++ texlive-bin-2009-modified/texk/kpathsea/readable.c	2011-03-13 16:44:41.000000000 +0100
@@ -58,9 +58,51 @@
 		  !(st & FILE_ATTRIBUTE_DIRECTORY));
 }
 #else
+
+#if 0
+/*
+ * This function here has the same functionality as the macro below. It was
+ * used to identify that the file not found error was due to the return
+ * value of stat as soley interpreted as a non existing file.
+ */
+static int
+readdable(const char * fn, struct stat *st)
+{
+  int flag;
+
+  flag = access(fn, R_OK);
+  if(flag != 0){
+    fprintf(stderr, "%s:%d readdable: Failed on access for %s\n",
+	    __FILE__, __LINE__,  fn);
+    return 0;
+  }
+  flag = stat (fn, st);
+  if(flag != 0){
+    fprintf(stderr, "%s:%d  readdable: Failed on stat for %s with errno %d ? EOVERFLOW = %d\n",
+	    __FILE__, __LINE__,  fn, errno, EOVERFLOW);
+    perror("Stat failed:");
+    return 0;
+  }
+
+  flag = S_ISDIR (st->st_mode);
+  if (flag == 0){
+    fprintf(stderr, "%s:%d  file readdable %s\n",
+	    __FILE__, __LINE__,  fn);
+      return 1;
+  }
+
+  fprintf(stderr, "%s:%d  readdable: Failed on stat for %s\n",
+	  __FILE__, __LINE__,  fn);
+  return 0;
+
+}
+#define  READABLE(fn, st) readdable(fn, &(st))
+#else
 #define READABLE(fn, st) \
   (access (fn, R_OK) == 0 && stat (fn, &(st)) == 0 && !S_ISDIR (st.st_mode))
 #endif
+#endif
+
 
 /* POSIX invented the brain-damage of not necessarily truncating
    filename components; the system's behavior is defined by the value of

--- End Message ---
--- Begin Message --- 
Am 13.03.2011 um 17:14 teilte Pierre SCHNIZER <p.schnizer@gsi.de> mit:

After going a little bit back and forward and some wild hacks we noticed
that it is a bit more complicated. Meanwhile:

1. the submitter is not interested any more and
2. the problem could have ben solved by migrating all computers from
32bit to 64bit.

The submitter agreed to close the issue.

Hilmar

>     thank you for all the effort putting a distriubtion together.
> 
> After upgrading to the lastly relased stable, I found that pdflatex would
> not find (and therefore compile) the files given on the command line,
> which were mounted by CIFS.
> 
> The error occurs in the kpathsea library (texk/kpathsea/readable.c),
> where the stat fails with the errno set to EOVERFLOW. It fails because
> the offset is to large. After I added manually -D_FILE_OFFSET_BITS=64
> (following the stat 2 man page), kpathsea/pdflatex finds the file and
> compiles it happily. I added "a patch", which illustrates my hack to
> find and circumvent the problem.
> 
> Could that be that AC_SYS_LARGEFILE is missing in configure.ac while it
> is listed in common.ac.orig?
> 
-- 
sigfault
#206401 http://counter.li.org

Attachment: signature.asc
Description: OpenPGP digital signature


--- End Message ---
Reply to: