nfs client problem in kernel 2.4.0
Hello,
I have a problem using the nfs client in kernel 2.4.0
My server is an SGI running IRIX 6.5.8
This tiny program, t1.c, displays the problem:
#include
#include
main()
{
DIR *dp;
struct dirent *de;
dp=opendir(".");
while((de=readdir(dp))!=NULL)
{
printf("%s\n", de->d_name);
}
closedir(dp);
}
It just simply lists the files in the current directory,
no big deal.
If I execute this on my redhat 7.0 machine running kernel 2.4.0,
in a directory exported from the SGI machine, one file is always
missing! This is not a permission problem, there's no problems
doing this from another machine running rh70 and kernel 2.2.16.
The output is:
$ ls -l
-rw-rw-r--1 mk carlsber 26 Jan 16 14:03 Makefile
-rwxrwxr-x1 mk carlsber13764 Jan 16 14:10 t1
-rw-rw-r--1 mk carlsber 194 Jan 16 14:10 t1.c
-rw-rw-r--1 mk carlsber 4509736 Jan 16 13:46 t1.dat
$ ./t1
.
..
t1
t1.c
t1.dat
The Makefile is not listed.
$ strace t1
open(".", O_RDONLY|O_NONBLOCK|O_DIRECTORY) = 3
fstat64(3, {st_mode=S_IFDIR|0775, st_size=65, ...}) = 0
fcntl64(3, F_SETFD, FD_CLOEXEC) = 0
brk(0) = 0x80496d8
brk(0x8059720) = 0x8059720
brk(0x805a000) = 0x805a000
getdents64(3, /* 6 entries */, 65536) = 160
lseek(3, 1547825467, SEEK_SET) = 1547825467
fstat64(1, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 0), ...}) = 0
old_mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS,
-1, 0) = 0x40018000
ioctl(1, TCGETS, {B38400 opost isig icanon echo ...}) = 0
write(1, ".\n", 2.
) = 2
write(1, "..\n", 3..
) = 3
write(1, "t1\n", 3t1
) = 3
write(1, "t1.c\n", 5t1.c
) = 5
write(1, "t1.dat\n", 7t1.dat
) = 7
getdents64(3, /* 1 entries */, 65536) = 32
close(3)= 0
munmap(0x40018000, 4096)= 0
_exit(0)= ?
Under kernel 2.2.16 I get the following from strace:
open(".", O_RDONLY|O_NONBLOCK|O_DIRECTORY) = 3
fstat64(3, 0xb630) = -1 ENOSYS (Function not
implemented)
fstat(3, {st_mode=S_IFDIR|0775, st_size=65, ...}) = 0
fcntl64(3, F_SETFD, FD_CLOEXEC) = -1 ENOSYS (Function not
implemented)
fcntl(3, F_SETFD, FD_CLOEXEC) = 0
brk(0) = 0x80496d8
brk(0x804a720) = 0x804a720
brk(0x804b000) = 0x804b000
getdents64(3, 0x8049710, 4096) = -1 ENOSYS (Function not
implemented)
getdents(3, /* 6 entries */, 3933) = 100
fstat(1, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 2), ...}) = 0
old_mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS,
-1, 0) = 0x40017000
ioctl(1, TCGETS, {B38400 opost isig icanon echo ...}) = 0
write(1, ".\n", 2.
) = 2
write(1, "..\n", 3..
) = 3
write(1, "t1\n", 3t1
) = 3
write(1, "t1.c\n", 5t1.c
) = 5
write(1, "t1.dat\n", 7t1.dat
) = 7
write(1, "Makefile\n", 9Makefile
) = 9
getdents(3, /* 0 entries */, 3933) = 0
close(3)= 0
munmap(0x40017000, 4096)= 0
_exit(0)= ?
Why do I see this difference?
BTW, I've tried both NFS2 and NFS3 (flag during compile
and nfsvers during mount), no difference.
If I use a Linux NFS server, the problem doesn't show up,
so it is the combination SGI NFS server and Linux 2.4.0
NFS client.
I first noticed this problem because one of the news servers I
use had disappeared from the list in netscape after upgrading to kernel
2.4.0,
and an strace lead me to suspect that readdir had a problem via
NFS to an SGI server.
Is this a kernel 2.4.0 problem or have I forgotten to upgrade
something else in rh70?
Mogens
--
Mogens Kjaer, Carlsberg Laboratory, Dept. of Chemistry
Gamle Carlsberg Vej 10, DK-2500 Valby, Denmark
Phone: +45 33 27 53 25, Fax: +45 33 27 47 08
Email: [EMAIL PROTECTED] Homepage: http://www.crc.dk
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
Please read the FAQ at http://www.tux.org/lkml/
Re: nfs client problem in kernel 2.4.0
Trond Myklebust wrote: > I'll bet it's the lseek that's screwing things up again. IIRC IRIX has > an export option to cause it to generate 32-bit readdir cookies. Could > you please try enabling it? Sorry, I forgot to mention this: This option was already enabled. Mogens -- Mogens Kjaer, Carlsberg Laboratory, Dept. of Chemistry Gamle Carlsberg Vej 10, DK-2500 Valby, Denmark Phone: +45 33 27 53 25, Fax: +45 33 27 47 08 Email: [EMAIL PROTECTED] Homepage: http://www.crc.dk - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] Please read the FAQ at http://www.tux.org/lkml/
Re: nfs client problem in kernel 2.4.0
Trond Myklebust wrote: > > >>>>> " " == Mogens Kjaer <[EMAIL PROTECTED]> writes: > >> getdents64(3, /* 6 entries */, 65536) = 160 lseek(3, > >> 1547825467, SEEK_SET) = 1547825467 ... getdents64(3, /* 1 > >> entries */, 65536) = 32 > > BTW: there does in any case seem to be a bug in your version of > glibc. getdents64() is returning 64-bit file offsets, so they're not > going to fit with ordinary lseek(). This turned out to be more difficult than I thought... I suspect glibc-2.2-12 being the reason, but I'm not quite sure yet: The problem is, that the 64-bit dirent's are converted to 32-bit dirent's and a sanity check is performed, if the inodes or offsets don't fit into 32 bits. The offset of the last entry is 4294967295 (no, not -1), this won't fit in a signed 32 bit number. Does this number come from the SGI or from the NFS stuff in the Linux kernel? Mogens -- Mogens Kjaer, Carlsberg Laboratory, Dept. of Chemistry Gamle Carlsberg Vej 10, DK-2500 Valby, Denmark Phone: +45 33 27 53 25, Fax: +45 33 27 47 08 Email: [EMAIL PROTECTED] Homepage: http://www.crc.dk - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] Please read the FAQ at http://www.tux.org/lkml/
Re: nfs client problem in kernel 2.4.0
Trond Myklebust wrote: > It comes from the SGI. The NFS client just considers it all a cookie, > and passes it on to glibc. We probably shouldn't do that, as indeed > the cookie is not guaranteed to be 32-bit signed, but it's what we > always did for 2.2.x. So what do I do to get it to work? I could patch glibc so that it treats the -1/4294967295 as a special case, but... (I actually did this, but updating /lib/libc.so on a running system turned out to be a really, Really, REALLY bad idea :-(( ). Before ruining my machine it worked by prepending LD_LIBRARY_PATH to the test program, so the idea works. Mogens -- Mogens Kjaer, Carlsberg Laboratory, Dept. of Chemistry Gamle Carlsberg Vej 10, DK-2500 Valby, Denmark Phone: +45 33 27 53 25, Fax: +45 33 27 47 08 Email: [EMAIL PROTECTED] Homepage: http://www.crc.dk - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] Please read the FAQ at http://www.tux.org/lkml/
Re: agpgart (2.4.0 kernel)
Robert Dale wrote: > > I have agpgart compiled into the kernel with both (Intel 440LX/BX/GX and > I815/I840/I850) and Intel I810/I815 drivers. When the kernel boots I get: > > Linux agpgart interface v0.99 (c) Jeff Hartmann > agpgart: Maximum main memory to use for agp memory: 27M > agppart: agpgart: Detected an Intel i815, but could not find > the secondary device. You can try one of the 2.4.0-ac kernels or look in the lkml archives for a threeline patch. Mogens -- Mogens Kjaer, Carlsberg Laboratory, Dept. of Chemistry Gamle Carlsberg Vej 10, DK-2500 Valby, Denmark Phone: +45 33 27 53 25, Fax: +45 33 27 47 08 Email: [EMAIL PROTECTED] Homepage: http://www.crc.dk - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] Please read the FAQ at http://www.tux.org/lkml/
Re: NFS Bug in 2.4.3?
"Thomas J. Baker" wrote:
>
> There is an NFS bug described here
>
> http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=30944
>
> that seems to have been known about for a while that is not fixed in
> 2.4.3. Is there something wrong with the patch that is discussed?
It works fine for me.
Note, however, the patch should be patched in order to work on the
2.4.2 kernel distributed with rh71.
--- linux-2.4.2-dir.dif Thu Mar 1 14:58:28 2001
+++ linux-2.4.2-dirRH71.dif Thu Apr 26 12:19:34 2001
@@ -111,13 +111,14 @@
diff -u --recursive --new-file linux-2.4.2-fh_align/fs/readdir.c
linux-2.4.2-dir/fs/readdir.c
--- linux-2.4.2-fh_align/fs/readdir.c Mon Dec 11 22:45:42 2000
+++ linux-2.4.2-dir/fs/readdir.c Thu Feb 22 10:47:49 2001
-@@ -315,7 +315,8 @@
+@@ -346,7 +346,9 @@
lastdirent = buf.previous;
if (lastdirent) {
struct linux_dirent64 d;
- d.d_off = file->f_pos;
++ /* d.d_off = file->f_pos; */
+ /* get the sign extension right */
+ d.d_off = (off_t)file->f_pos;
- copy_to_user(&lastdirent->d_off, &d.d_off,
sizeof(d.d_off));
error = count - buf.count;
- }
+ if (copy_to_user(&lastdirent->d_off, &d.d_off,
sizeof(d.d_off)))
+ error = -EFAULT;
(patch to a patch; hm...).
Mogens
--
Mogens Kjaer, Carlsberg Laboratory, Dept. of Chemistry
Gamle Carlsberg Vej 10, DK-2500 Valby, Denmark
Phone: +45 33 27 53 25, Fax: +45 33 27 47 08
Email: [EMAIL PROTECTED] Homepage: http://www.crc.dk
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
