Custom Query (16363 matches)
Results (1654 - 1656 of 16363)
| Ticket | Resolution | Summary | Owner | Reporter |
|---|---|---|---|---|
| #2921 | fixed | sendfile => Fixed in SVN | ||
| Description |
Hi, Using Perlbal as a webserver with a document root hosted on a vboxsf mount share, I noticed that no content was sent to the socket, Perlbal was erroring with the following message: "Error w/ sendfile: Value too large for defined data type" I believe Perlbal is not the only one impacted by this issue (might be a regression of ticket #819?) Perlbal has some external dependencies (Sys::Syscall, IO::AIO etc...) that I tried to remove for you. Using perl 5.10 of ubuntu 32bits and the following test script reproduces the error: #!/usr/bin/perl
use strict;
use warnings;
use Config;
use POSIX;
use IO::Socket::INET;
my $src = shift;
open (my $s, "<$src") or die "cannot open $src";
my $d = open_sock();
doit($s, $d);
sub doit {
my ($fhs, $fhd) = @_;
my $src_fd = fileno($fhs);
my $dst_fd = fileno($fhd);
my $size = -s $fhs;
my $SYS_sendfile = 187;
my $machine = (POSIX::uname())[4];
unless ($^O eq "linux") {
die "I don't know if not under linux";
}
if ($machine eq "x86_64" && $Config{ptrsize} == 4) {
$machine = "i386";
}
unless ($machine =~ m/^i[3456]86$/) {
die "I can't tell for $machine";
}
my $res = syscall(
$SYS_sendfile,
$dst_fd,
$src_fd,
0,
$size,
);
if ($res < 0) {
warn "FAIL: $!";
}
else {
print "SUCCESS\n";
}
}
sub open_sock {
my $port = shift || 9245;
print "now connect to localhost:$port, thanks\n";
my $listen = IO::Socket::INET->new(
Listen => 1,
LocalAddr => 'localhost',
LocalPort => $port,
Proto => 'tcp',
Timeout => 15,
ReuseAddr => 1,
);
die "problem $!" unless $listen;
return $listen->accept or die "accept failed";
}
you will need to:
Broken case: when the filepath is in the vboxsf mount: yann@dev:~/dev$ perl repro.pl /in/vboxsf now connect to localhost:9245, thanks FAIL: Value too large for defined data type at repro2.pl line 40. Normal case: when the filepath is in the normal filesystem: yann@dev:~/dev$ perl repro.pl /tmp/src now connect to localhost:9245, thanks SUCCESS here is a strace() of the failing system call: sendfile(5, 3, NULL, 1497) = -1 EOVERFLOW (Value too large for defined data type) write(2, "FAIL: Value too large for define"..., 66FAIL: Value too large for defined data type at repro.pl line 39. ) = 66 here is my mount: $ mount -t vboxsf devel on /home/yann/dev type vboxsf (uid=1000,gid=1000,rw) $ uname -a Linux dev 2.6.27-9-server #1 SMP Thu Nov 20 22:53:41 UTC 2008 i686 GNU/Linux Thanks! Yann Kerherve |
|||
| #2924 | fixed | Red Hat ES3 64bit does not run on VB 2.0.6 | ||
| Description |
I have installed Red Hat ES3 on a VirtualBox 2.0.6 running on a AMD 64bit. I tried Solaris 10 x64 and Vista Business x64 as host system, but both show the same behaviour. The installation works fine, but when I try to boot the guest I get a constant stream of error messages: "hda: DMA interrupt recovery" and "hda: lost interrupt" Due these messages the system takes hours to boot and is virtually unusable. I did a complete system update (which takes about 3 days) but the error messages still appear. I tried to install the VBox Additions but that fails. I raised a support case with Red Hat (Service Request: 1874786) and they confirmed this behaviour. I have no problems running RHEL3 32bit and RHEL5 64bit. |
|||
| #2925 | fixed | Guru Meditation on Windows XP guest running VBox 2.1.0 on MacOS 10.5.6 -> fixed in SVN | ||
| Description |
Am receiving a guru meditation on 2.1.0 on macos 10.5.6. XP guest will not stay running for more than 24 hours. Both instances have happened while I was away from the computer and was unable to determine what was going on. Have attached the VBox.log file as well as the VBox.png file. |
|||

