* [Bug:cli] Loading user-defined function generates an internal error
@ 2009-12-30 5:35 Nick Roberts
2009-12-30 5:46 ` Joel Brobecker
0 siblings, 1 reply; 9+ messages in thread
From: Nick Roberts @ 2009-12-30 5:35 UTC (permalink / raw)
To: gdb
If the user-defined function below is put in a file user.cmd, say, then in
internal error is generated upon loading it:
..
(gdb) source user.cmd
utils.c:1206: internal-error: virtual memory exhausted.
A problem internal to GDB has been detected,
further debugging may prove unreliable.
Quit this debugging session? (y or n) [answered Y; input not from terminal]
utils.c:1206: internal-error: virtual memory exhausted.
A problem internal to GDB has been detected,
further debugging may prove unreliable.
Create a core file of GDB? (y or n) [answered Y; input not from terminal]
Aborted (core dumped)
This didn't used to happen, say 3 months ago.
--
Nick http://users.snap.net.nz/~nickrob
define p_tree
set $tree = ($arg0)
set $i = 0
set $node = $tree->_M_t->_M_impl->_M_header->_M_left
set $tree_size = $tree->_M_t->_M_impl->_M_node_count
while ($i < $tree_size)
set $i++
printf "NODE %d: ", $i
set $value = (void *)($node + 1)
p *($arg1 *)$value
set $value = $value + 4
p *($arg2 *)$value
# Next (bigger) value must be to right...
if ($node->_M_right != 0)
set $node = $node->_M_right
# Descend tree while there is a left node.
while ($node->_M_left != 0)
set $node = $node->_M_left
end
# ...or further up the tree.
else
set $tmp_node = $node->_M_parent
# Ascend tree while at right node.
while ($node == $tmp_node->_M_right)
set $node = $tmp_node
set $tmp_node = $tmp_node->_M_parent
end
# Set to parent of first left node (condition always true?)
if ($node->_M_right != $tmp_node)
set $node = $tmp_node
end
end
end
end
^ permalink raw reply [flat|nested] 9+ messages in thread* Re: [Bug:cli] Loading user-defined function generates an internal error
2009-12-30 5:35 [Bug:cli] Loading user-defined function generates an internal error Nick Roberts
@ 2009-12-30 5:46 ` Joel Brobecker
2009-12-30 8:50 ` Nick Roberts
2009-12-30 22:26 ` Nick Roberts
0 siblings, 2 replies; 9+ messages in thread
From: Joel Brobecker @ 2009-12-30 5:46 UTC (permalink / raw)
To: Nick Roberts; +Cc: gdb
> If the user-defined function below is put in a file user.cmd, say, then in
> internal error is generated upon loading it:
I can reproduce too. Have you tried debugging the problem?
--
Joel
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [Bug:cli] Loading user-defined function generates an internal error
2009-12-30 5:46 ` Joel Brobecker
@ 2009-12-30 8:50 ` Nick Roberts
2009-12-31 6:27 ` Vladimir Prus
2009-12-30 22:26 ` Nick Roberts
1 sibling, 1 reply; 9+ messages in thread
From: Nick Roberts @ 2009-12-30 8:50 UTC (permalink / raw)
To: Joel Brobecker; +Cc: gdb
> > If the user-defined function below is put in a file user.cmd, say, then in
> > internal error is generated upon loading it:
>
> I can reproduce too. Have you tried debugging the problem?
It's due to this change:
2009-11-18 Vladimir Prus <vladimir@codesourcery.com>
* cli/cli-script.c (process_next_line): Recognize 'end'
even when the line has leading space and we're not parsing
commands.
--
Nick http://users.snap.net.nz/~nickrob
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [Bug:cli] Loading user-defined function generates an internal error
2009-12-30 8:50 ` Nick Roberts
@ 2009-12-31 6:27 ` Vladimir Prus
0 siblings, 0 replies; 9+ messages in thread
From: Vladimir Prus @ 2009-12-31 6:27 UTC (permalink / raw)
To: gdb
Nick Roberts wrote:
> > > If the user-defined function below is put in a file user.cmd, say, then in
> > > internal error is generated upon loading it:
> >
> > I can reproduce too. Have you tried debugging the problem?
>
> It's due to this change:
>
> 2009-11-18 Vladimir Prus <vladimir@codesourcery.com>
>
> * cli/cli-script.c (process_next_line): Recognize 'end'
> even when the line has leading space and we're not parsing
> commands.
Do you happen to know what is wrong, exactly? Unfortunately, I won't
be able to address this until Jan 11, so if you don't have the time
to investigate further, I suggest you open an issue in bugzilla.
Thanks,
Volodya
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [Bug:cli] Loading user-defined function generates an internal error
2009-12-30 5:46 ` Joel Brobecker
2009-12-30 8:50 ` Nick Roberts
@ 2009-12-30 22:26 ` Nick Roberts
2010-01-01 8:41 ` [patch] Fix crash on CLI indented comments [Re: [Bug:cli] Loading user-defined function generates an internal error] Jan Kratochvil
1 sibling, 1 reply; 9+ messages in thread
From: Nick Roberts @ 2009-12-30 22:26 UTC (permalink / raw)
To: Joel Brobecker; +Cc: gdb
> I can reproduce too. Have you tried debugging the problem?
Heres a simpler user defined function that gives a segmentation fault. It
appears that GDB can no longer handle indented comments.
define my_fun
#indented comment
end
--
Nick http://users.snap.net.nz/~nickrob
^ permalink raw reply [flat|nested] 9+ messages in thread
* [patch] Fix crash on CLI indented comments [Re: [Bug:cli] Loading user-defined function generates an internal error]
2009-12-30 22:26 ` Nick Roberts
@ 2010-01-01 8:41 ` Jan Kratochvil
2010-01-01 10:36 ` Joel Brobecker
0 siblings, 1 reply; 9+ messages in thread
From: Jan Kratochvil @ 2010-01-01 8:41 UTC (permalink / raw)
To: Nick Roberts; +Cc: Joel Brobecker, gdb, gdb-patches
On Wed, 30 Dec 2009 23:26:27 +0100, Nick Roberts wrote:
> Heres a simpler user defined function that gives a segmentation fault. It
> appears that GDB can no longer handle indented comments.
>
> define my_fun
> #indented comment
> end
(*command)->line = savestring (p, p1 - p);
#1 0x0000000000487405 in savestring (ptr=0x1d5dd31 "", size=18446744073709551615) at utils.c:1377
(gdb) p/x size
$1 = 0xffffffffffffffff
(gdb) p p
$2 = 0x1d5dd31 ""
(gdb) p p1
$3 = 0x1d5dd30 " "
OK to check-in? Probably [obv].
Thanks,
Jan
2010-01-01 Jan Kratochvil <jan.kratochvil@redhat.com>
* cli/cli-script.c (process_next_line): Check P2 overrun.
--- a/gdb/cli/cli-script.c
+++ b/gdb/cli/cli-script.c
@@ -893,7 +893,7 @@ process_next_line (char *p, struct command_line **command, int parse_commands)
p2 = p;
/* Strip leading whitespace. */
- while (*p2 == ' ' || *p2 == '\t')
+ while (p2 != p1 && (*p2 == ' ' || *p2 == '\t'))
p2++;
/* 'end' is always recognized, regardless of parse_commands value.
^ permalink raw reply [flat|nested] 9+ messages in thread* Re: [patch] Fix crash on CLI indented comments [Re: [Bug:cli] Loading user-defined function generates an internal error]
2010-01-01 8:41 ` [patch] Fix crash on CLI indented comments [Re: [Bug:cli] Loading user-defined function generates an internal error] Jan Kratochvil
@ 2010-01-01 10:36 ` Joel Brobecker
2010-01-01 11:01 ` Jan Kratochvil
2010-01-01 11:12 ` Joel Brobecker
0 siblings, 2 replies; 9+ messages in thread
From: Joel Brobecker @ 2010-01-01 10:36 UTC (permalink / raw)
To: Jan Kratochvil; +Cc: Nick Roberts, gdb, gdb-patches
[-- Attachment #1: Type: text/plain, Size: 642 bytes --]
> 2010-01-01 Jan Kratochvil <jan.kratochvil@redhat.com>
>
> * cli/cli-script.c (process_next_line): Check P2 overrun.
Yes - I agree it's obvious when you know that the comment is stripped
upstream...
Approved. Thanks for fixing.
Given the number of regressions/failures we've seen in this area,
I really thought we should add a test for this, so I created
the attached patch. I will commit after you have committed yours.
gdb/testsuite/
Test indented comment in file being sourced.
* commands.exp: Test indented comment in file being sourced.
Tested on x86_64-linux. Fails miserably without Jan's patch.
--
Joel
[-- Attachment #2: test-indented-comment.diff --]
[-- Type: text/x-diff, Size: 1246 bytes --]
commit 0f79c42a551762d87088fd5916c4ed390dde9942
Author: Joel Brobecker <brobecker@adacore.com>
Date: Fri Jan 1 14:31:07 2010 +0400
Test indented comment in file being sourced.
* commands.exp: Test indented comment in file being sourced.
diff --git a/gdb/testsuite/gdb.base/commands.exp b/gdb/testsuite/gdb.base/commands.exp
index 2aba51f..228c464 100644
--- a/gdb/testsuite/gdb.base/commands.exp
+++ b/gdb/testsuite/gdb.base/commands.exp
@@ -577,6 +577,19 @@ proc stray_arg0_test { } {
"stray_arg0_test #4"
}
+# Test that GDB is able to source a file with an indented comment.
+proc source_file_with_indented_comment {} {
+ set fd [open "file1" w]
+ puts $fd \
+{define my_fun
+ #indented comment
+end
+echo Done!\n}
+ close $fd
+
+ gdb_test "source file1" "Done!" "source file with indented comment"
+}
+
# Test that GDB can handle arguments when sourcing files recursively.
# If the arguments are overwritten with ####### then the test has failed.
proc recursive_source_test {} {
@@ -761,6 +774,7 @@ deprecated_command_test
bp_deleted_in_command_test
temporary_breakpoint_commands
stray_arg0_test
+source_file_with_indented_comment
recursive_source_test
if_commands_test
redefine_hook_test
^ permalink raw reply [flat|nested] 9+ messages in thread* Re: [patch] Fix crash on CLI indented comments [Re: [Bug:cli] Loading user-defined function generates an internal error]
2010-01-01 10:36 ` Joel Brobecker
@ 2010-01-01 11:01 ` Jan Kratochvil
2010-01-01 11:12 ` Joel Brobecker
1 sibling, 0 replies; 9+ messages in thread
From: Jan Kratochvil @ 2010-01-01 11:01 UTC (permalink / raw)
To: Joel Brobecker; +Cc: Nick Roberts, gdb, gdb-patches
On Fri, 01 Jan 2010 11:35:20 +0100, Joel Brobecker wrote:
> Approved. Thanks for fixing.
Checked-in.
> Test indented comment in file being sourced.
> * commands.exp: Test indented comment in file being sourced.
OK, thanks for this work.
Regards,
Jan
http://sourceware.org/ml/gdb-cvs/2010-01/msg00014.html
--- src/gdb/ChangeLog 2010/01/01 09:44:05 1.11190
+++ src/gdb/ChangeLog 2010/01/01 10:57:42 1.11191
@@ -1,3 +1,7 @@
+2010-01-01 Jan Kratochvil <jan.kratochvil@redhat.com>
+
+ * cli/cli-script.c (process_next_line): Check P2 overrun.
+
2009-01-01 Joel Brobecker <brobecker@adacore.com>
Update the copyright hearder to add year 2010 for most GDB files.
--- src/gdb/cli/cli-script.c 2010/01/01 07:31:47 1.55
+++ src/gdb/cli/cli-script.c 2010/01/01 10:57:43 1.56
@@ -893,7 +893,7 @@
p2 = p;
/* Strip leading whitespace. */
- while (*p2 == ' ' || *p2 == '\t')
+ while (p2 != p1 && (*p2 == ' ' || *p2 == '\t'))
p2++;
/* 'end' is always recognized, regardless of parse_commands value.
^ permalink raw reply [flat|nested] 9+ messages in thread* Re: [patch] Fix crash on CLI indented comments [Re: [Bug:cli] Loading user-defined function generates an internal error]
2010-01-01 10:36 ` Joel Brobecker
2010-01-01 11:01 ` Jan Kratochvil
@ 2010-01-01 11:12 ` Joel Brobecker
1 sibling, 0 replies; 9+ messages in thread
From: Joel Brobecker @ 2010-01-01 11:12 UTC (permalink / raw)
To: Jan Kratochvil; +Cc: Nick Roberts, gdb, gdb-patches
> gdb/testsuite/
>
> Test indented comment in file being sourced.
> * gdb.base/commands.exp: Test indented comment in file being sourced.
This patch is now in as well. (notice that I added the missing gdb.base
in the name of the file - oops)
--
Joel
^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2010-01-01 11:12 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2009-12-30 5:35 [Bug:cli] Loading user-defined function generates an internal error Nick Roberts
2009-12-30 5:46 ` Joel Brobecker
2009-12-30 8:50 ` Nick Roberts
2009-12-31 6:27 ` Vladimir Prus
2009-12-30 22:26 ` Nick Roberts
2010-01-01 8:41 ` [patch] Fix crash on CLI indented comments [Re: [Bug:cli] Loading user-defined function generates an internal error] Jan Kratochvil
2010-01-01 10:36 ` Joel Brobecker
2010-01-01 11:01 ` Jan Kratochvil
2010-01-01 11:12 ` Joel Brobecker
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox