[Bug 1062576] New: memory leak when including a file with "use utf8"
by Red Hat Bugzilla
https://bugzilla.redhat.com/show_bug.cgi?id=1062576
Bug ID: 1062576
Summary: memory leak when including a file with "use utf8"
Product: Fedora
Version: 19
Component: perl
Severity: medium
Assignee: jplesnik(a)redhat.com
Reporter: lav(a)yars.free.net
QA Contact: extras-qa(a)fedoraproject.org
CC: cweyl(a)alumni.drew.edu, iarnell(a)gmail.com,
jplesnik(a)redhat.com, kasal(a)ucw.cz,
perl-devel(a)lists.fedoraproject.org, ppisar(a)redhat.com,
psabata(a)redhat.com, rc040203(a)freenet.de,
tcallawa(a)redhat.com
Description of problem:
Every time perl includes a file with "use utf8" and text constants it leaks
memory.
Version-Release number of selected component (if applicable):
perl-5.16.3-266.fc19.x86_64
How reproducible:
always
Steps to Reproduce:
1. run the test program below
2. watch as memory usage grows
Here is the test program:
-------------------------
my $mem = 0;
sub report_memory () {
my $next_mem = qx[ps -p $$ -o size=]+0;
printf "%+8d = %8d K\n", $next_mem-$mem, $next_mem; $mem = $next_mem;
}
for (0..1e5) {
do 'x.inc';
report_memory unless $_ % 1e4;
}
-------------------------
x.inc contains:
-------------------------
use utf8;
$x='x';
-------------------------
--
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=l7ghbj1UQE&a=cc_unsubscribe
8 years, 6 months
[Bug 1129850] New: Perl debugger commands that depend on Padwalker no longer work
by Red Hat Bugzilla
https://bugzilla.redhat.com/show_bug.cgi?id=1129850
Bug ID: 1129850
Summary: Perl debugger commands that depend on Padwalker no
longer work
Product: Fedora
Version: 20
Component: perl
Severity: high
Assignee: jplesnik(a)redhat.com
Reporter: alan(a)clueserver.org
QA Contact: extras-qa(a)fedoraproject.org
CC: cweyl(a)alumni.drew.edu, iarnell(a)gmail.com,
jplesnik(a)redhat.com, kasal(a)ucw.cz,
perl-devel(a)lists.fedoraproject.org, ppisar(a)redhat.com,
psabata(a)redhat.com, rc040203(a)freenet.de,
tcallawa(a)redhat.com
Description of problem:
When using the Perl command line debugger the "y" and "X" commands no longer
work. (They fail silently.)
Version-Release number of selected component (if applicable):
perl-5.18.2-289.fc20
perl-PadWalker-1.98-1.fc20.x86_64
How reproducible:
Every time.
Steps to Reproduce:
1. Debug a test program with "perl -d testprog.pl
2. Use "n" and enter a few lines until a variable you want to examine is
loaded.
3. Use "y" or "X" on that variable.
Actual results:
Nothing returned.
Expected results:
The contents of the variable should be displayed.
Additional info:
The "p" command will print a variable, it is very narrow in scope. It will not
display the contents of an entire hash variable like "X" or "y" will. ("y"
shows the contents of the variable for all types in the current package. "X"
does the same for global variables. Neither work at this point.
--
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=VpjaTqvTg7&a=cc_unsubscribe
8 years, 6 months
[Bug 839612] New: FTBFS perl-Unix-Statgrab-0.04-14.fc18
by Red Hat Bugzilla
https://bugzilla.redhat.com/show_bug.cgi?id=839612
Bug ID: 839612
QA Contact: extras-qa(a)fedoraproject.org
Severity: unspecified
URL: https://koji.fedoraproject.org/koji/taskinfo?taskID=42
36283
Version: rawhide
Priority: unspecified
CC: oliver(a)linux-kernel.at,
perl-devel(a)lists.fedoraproject.org, steve(a)silug.org
Assignee: steve(a)silug.org
Summary: FTBFS perl-Unix-Statgrab-0.04-14.fc18
Regression: ---
Story Points: ---
Classification: Fedora
OS: Unspecified
Reporter: ppisar(a)redhat.com
Type: Bug
Documentation: ---
Hardware: Unspecified
Mount Type: ---
Status: NEW
Component: perl-Unix-Statgrab
Product: Fedora
perl-Unix-Statgrab-0.04-14.fc18 does not build in F18:
Executing(%check): /bin/sh -e /var/tmp/rpm-tmp.FI62R4
+ umask 022
+ cd /builddir/build/BUILD
+ cd Unix-Statgrab-0.04
+ unset DISPLAY
+ make test
PERL_DL_NONLAZY=1 /usr/bin/perl "-MExtUtils::Command::MM" "-e" "test_harness(0,
'blib/lib', 'blib/arch')" t/*.t
t/0_pod.t ........... ok
t/0_pod_coverage.t .. ok
t/Unix-Statgrab.t ...
Failed 1/22 subtests
Test Summary Report
-------------------
t/Unix-Statgrab.t (Wstat: 11 Tests: 21 Failed: 0)
Non-zero wait status: 11
Parse errors: Bad plan. You planned 22 tests but ran 21.
Files=3, Tests=23, 0 wallclock secs ( 0.04 usr 0.00 sys + 0.15 cusr 0.03
csys = 0.22 CPU)
Result: FAIL
Failed 1/3 test programs. 0/23 subtests failed.
make: *** [test_dynamic] Error 255
This happens with perl 5.16.0. I guess the test plan is wrong or the perl
segfaults.
--
You are receiving this mail because:
You are on the CC list for the bug.
8 years, 7 months
[Bug 750805] New: Fails to build on ARM, needs to use default setjmp not ucontext
by Red Hat Bugzilla
Please do not reply directly to this email. All additional
comments should be made in the comments box of this bug.
Summary: Fails to build on ARM, needs to use default setjmp not ucontext
https://bugzilla.redhat.com/show_bug.cgi?id=750805
Summary: Fails to build on ARM, needs to use default setjmp not
ucontext
Product: Fedora
Version: 15
Platform: Unspecified
OS/Version: Unspecified
Status: NEW
Severity: unspecified
Priority: unspecified
Component: perl-Coro
AssignedTo: ppisar(a)redhat.com
ReportedBy: henrik(a)henriknordstrom.net
QAContact: extras-qa(a)fedoraproject.org
CC: fedora-perl-devel-list(a)redhat.com,
mmaslano(a)redhat.com, kwizart(a)gmail.com,
bochecha(a)fedoraproject.org, ppisar(a)redhat.com
Classification: Fedora
Story Points: ---
Type: ---
Description of problem:
ARM do not implement the needed ucontext functions, and need to use the default
setjmp method. This is normally the default, except that fedora patches it to
hardwire ucontext as default method..
Version-Release number of selected component (if applicable):
perl-Coro-5.372-3.fc15
How reproducible:
always
Steps to Reproduce:
1. Try to rebuild perl-Coro on arm
2.
3.
Actual results:
failed build, crashing in testsuite
Expected results:
successful build
Additional info:
Trivial spec file patch attached.
Please apply patch and sumbit a F15 koji build from which arm can pull the
srpm. update request is not stricty needed if the only change relative to
current F15 build is this patch.
--
Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.
8 years, 9 months
[Bug 1068706] New: Password prompt matching requires colon, which may be missing
by Red Hat Bugzilla
https://bugzilla.redhat.com/show_bug.cgi?id=1068706
Bug ID: 1068706
Summary: Password prompt matching requires colon, which may be
missing
Product: Fedora
Version: 20
Component: perl-Net-SFTP-Foreign
Severity: low
Assignee: fedorapkg(a)rule.lv
Reporter: ejtr(a)layer3.co.uk
QA Contact: extras-qa(a)fedoraproject.org
CC: fedorapkg(a)rule.lv, perl-devel(a)lists.fedoraproject.org
Description of problem:
Net::SFTP::Foreign's password prompt matching code goes into a permanent loop
when the remote server supplies a prompt without a trailing colon.
Version-Release number of selected component (if applicable):
perl-Net-SFTP-Foreign-1.75-3.fc20.noarch.rpm
How reproducible:
When testing Net::SFTP::Foreign against an "unusual" host which supported
keyboard-interactive authentication, and thus generated a server-supplied
password prompt, it was found that the host in question had a prompt of the
form "username's password", rather than the more usual "username@host's
password: " as seen from OpenSSH client code supporting the password
authentication mode.
As a consequence, the Net::SFTP::Foreign module simply sulked waiting for a
colon.
Looking at the relevant
/usr/share/perl5/vendor_perl/Net/SFTP/Foreign/Backend/Unix.pm code file, we
find the following code fragment:
.....
debug and $debug & 65536 and _debug "looking for user/password prompt";
my $re = ( defined $password_prompt
? $password_prompt
: qr/(user|name|login)?[:?]\s*$/i );
$debug and $debug & 65536 and _debug "matching against $re";
if (substr($buffer, $at) =~ $re) {
if ($ask_for_username_at_login and
($ask_for_username_at_login ne 'auto' or defined $1)) {
$debug and $debug & 65536 and _debug "sending username";
print $pty "$user\n";
undef $ask_for_username_at_login;
}
else {
$debug and $debug & 65536 and _debug "sending password";
print $pty "$pass\n";
$password_sent = 1;
}
$at = length $buffer;
}
.....
In the absence of the $password_prompt setting, $re only matches a password
prompt containing the string "password" if it contains a colon or question-mark
after the word "password".
The code is also questionable in being capable of sending a password in
response to a prompt which contains only a colon followed by whitespace; I feel
it ought to at least see the word "password" or "passphrase".
The code also uses the $password_prompt variable to match against a potential
username prompt, which is questionable.
I believe it would be better to more completely separate the user and password
prompt matching with different Regex's, and avoid being able to match against
the short ":" string alone, and cope with the unusual condition of a missing
colon.
This recoding of the relevant block is a potential solution:
.....
$debug and $debug & 65536 and _debug "looking for user/password prompt";
my $reuser = qr/(user|name|login)([:?])?\s*$/i;
my $repass = ( defined $password_prompt ?
$password_prompt : qr/(password|passphrase)([:?])?\s*$/i );
if (substr($buffer, $at) =~ $reuser) {
$debug and $debug & 65536 and _debug "matched against $reuser";
if ($ask_for_username_at_login and
($ask_for_username_at_login ne 'auto' or defined $1)) {
$debug and $debug & 65536 and _debug "sending username";
print $pty "$user\n";
undef $ask_for_username_at_login;
}
$at = length $buffer;
} elsif (substr($buffer, $at) =~ $repass) {
$debug and $debug & 65536 and _debug "matched against $repass";
$debug and $debug & 65536 and _debug "sending password";
print $pty "$pass\n";
$password_sent = 1;
$at = length $buffer;
}
....
--
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=SXtN9EENTL&a=cc_unsubscribe
8 years, 9 months
[Bug 1135624] New: perl-Clipboard: insecure temporary file usage
by Red Hat Bugzilla
https://bugzilla.redhat.com/show_bug.cgi?id=1135624
Bug ID: 1135624
Summary: perl-Clipboard: insecure temporary file usage
Product: Security Response
Component: vulnerability
Keywords: Security
Severity: low
Priority: low
Assignee: security-response-team(a)redhat.com
Reporter: vdanen(a)redhat.com
CC: iarnell(a)gmail.com, mkreder(a)gmail.com,
perl-devel(a)lists.fedoraproject.org
It was reported [1],[2] that the clipedit program as shipped with
perl-Clipboard uses temporary files insecurely (based on the PID of the running
program). Using symlink attacks, an attacker could cause the deletion of
arbitrary files that the user running clipedit has write access to.
[...]
7 my $tmpfilename = "/tmp/clipedit$$";
8 open my $tmpfile, ">$tmpfilename" or die "Failure to open $tmpfilename:
$!";
9 print $tmpfile $orig;
10 close $tmpfile;
[...]
13 system($ed, $tmpfilename);
14
15 open $tmpfile, $tmpfilename or die "Failure to open $tmpfilename: $!";
16 my $edited = join '', <$tmpfile>;
[...]
49 unlink($tmpfilename) or die "Couldn't remove $tmpfilename: $!";
[1] http://seclists.org/oss-sec/2014/q3/467
[2] https://rt.cpan.org/Public/Bug/Display.html?id=98435
--
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=71pk61Zg6T&a=cc_unsubscribe
8 years, 9 months