Anyone else being hit by this?
,
| ...
| .debug_loc
| *(.debug_loc)
|
| .debug_macinfo
| *(.debug_macinfo)
|
| .debug_weaknames
| *(.debug_weaknames)
|
| .debug_funcnames
| *(.debug_funcnames)
|
| .debug_typenames
| *(.debug_typenames)
|
| .debug_varnames
| *(.debug_varnames)
|
|
On Mon, 7 May 2007, Denis Vlasenko wrote:
On Monday 07 May 2007 17:27, Cristian Ionescu-Idbohrn wrote:
Anyone else being hit by this?
,
| ...
| .debug_varnames
| *(.debug_varnames)
|
| /DISCARD/
| *(.note.GNU-stack)
| OUTPUT(busybox_unstripped elf32-i386)
| collect2
Trunk bb-build fails with:
findutils/lib.a(find.o): In function `parse_params':
find.c:(.text.parse_params+0x1b0): undefined reference to `xregcomp'
find.c:(.text.parse_params+0x201): undefined reference to `plus_minus_num'
if:
# CONFIG_FEATURE_FIND_MTIME is not set
# CONFIG_FEATURE_FIND_MMIN
Cheers,
--
CristianIndex: findutils/find.c
===
--- findutils/find.c (revision 18980)
+++ findutils/find.c (working copy)
@@ -75,8 +75,8 @@
} action;
#define ACTS(name, arg...) typedef struct { action a; arg; } action_##name;
On Sun, 1 Jul 2007, Denis Vlasenko wrote:
On Sunday 01 July 2007 18:08, Cristian Ionescu-Idbohrn wrote:
Fix warning:
procps/ps.c: In function `ps_main':
procps/ps.c:343: warning: unused variable `i'
I just made some changes to ps in svn, should work now.
Yell if it does not.
Looks
findutils/find.c: In function `parse_params':
findutils/find.c:512: warning: no previous prototype for `alloc_action'
Cheers,
--
Cristian
___
busybox mailing list
busybox@busybox.net
http://busybox.net/cgi-bin/mailman/listinfo/busybox
On Sun, 1 Jul 2007, Denis Vlasenko wrote:
Do you mean that your gcc emits bogus warning for l too?
Yes.
This is better (no actual code generated:
VALUE *l = l; /* silence gcc */
VALUE *v = v; /* silence gcc */
It's fine with me, as long as it makes gcc shut up (which it does).
...to skip warnings:
.config:75:warning: symbol value '' invalid for FEATURE_EDITING_MAX_LEN
.config:78:warning: symbol value '' invalid for FEATURE_EDITING_HISTORY
.config:300:warning: symbol value '' invalid for FEATURE_VI_MAX_LEN
.config:498:warning: symbol value '' invalid for
On Sun, 1 Jul 2007, Denis Vlasenko wrote:
On Sunday 01 July 2007 19:25, Cristian Ionescu-Idbohrn wrote:
findutils/find.c: In function `parse_params':
findutils/find.c:512: warning: no previous prototype for `alloc_action'
I think it's possible only thru gcc bugzilla.
Let's hope so
On Wed, 4 Jul 2007, Jim Freeman wrote:
On Wed, Jul 04, 2007 at 05:39:25PM +0200, Cristian Ionescu-Idbohrn wrote:
On Tue, 3 Jul 2007, Jim Freeman wrote:
# passwd -p blip
Isn't this the well known insecure method that shouldn't be used
because (with the right timing) anyone can
Had some troubles finding the expressions that work the same way in
both bash, dash and ash (various versions) and put together a small
script to help with the exercise. Couldn't find anything similar in
the testsuite dir, so I thought it might be useful.
Cheers,
--
Cristian
AFAICS, that option creates a temporary file on the same fs as the edited
file resides on. And that is indeed optimal, but at the same time
somewhat unfortunate for embedded systems.
Embedded systems usualy place editable files on flash fs and those fs get
used out very much faster compared to
On Sun, 26 Aug 2007, Denys Vlasenko wrote:
On Saturday 25 August 2007 23:10, Cristian Ionescu-Idbohrn wrote:
AFAICS, that option creates a temporary file on the same fs as the edited
file resides on. And that is indeed optimal, but at the same time
somewhat unfortunate for embedded
On Sun, 26 Aug 2007, Denys Vlasenko wrote:
On Sunday 26 August 2007 16:00, Cristian Ionescu-Idbohrn wrote:
Embedded systems usualy place editable files on flash fs and those
fs get used out very much faster compared to real hdfs.
I think that sed creates new file, fills
BusyBox v1.8.1 (2007-11-12 19:47:38 CET) multi-call binary
/var/empty/ is a directory.
# tr '\0' ' ' /var/empty/
Segmentation fault
Can anyone reproduce this?
Cheers,
--
Cristian
___
busybox mailing list
busybox@busybox.net
On Tue, 13 Nov 2007, Cristian Ionescu-Idbohrn wrote:
Date: Tue, 13 Nov 2007 16:04:53 +0100 (CET)
From: Cristian Ionescu-Idbohrn [EMAIL PROTECTED]
Reply-To: busybox@busybox.net
To: busybox@busybox.net
Subject: tr segfaults
BusyBox v1.8.1 (2007-11-12 19:47:38 CET) multi-call binary
/var
On Tue, 13 Nov 2007, Denys Vlasenko wrote:
On Monday 12 November 2007 08:51, Peter Kjellerstedt wrote:
Denis,
older versions of gcc do not support the -Wl,--sort-section
option. The following patch to scripts/trylink handles this.
Added to svn, thanks!
On Tue, 13 Nov 2007, Denys Vlasenko wrote:
On Tuesday 13 November 2007 09:37, Fernando Silveira wrote:
Yep, but the patch follows inline. The read(2) return value was being
stored in an unsigned variable, avoiding the error detection. I also
added an error message, which wasn't
On Fri, 16 Nov 2007, Denys Vlasenko wrote:
I updated tar fix:
http://busybox.net/downloads/fixes-1.8.1/busybox-1.8.1-tar_z.patch
Denys,
Would it be possible, please, to put incremental/numbered patches in that
directory instead. Updated patches are much more difficult to track.
Cheers,
On Thu, 22 Nov 2007, Denys Vlasenko wrote:
About the -W seconds. I added it because of what was pointed out in the
RFC2131:
5. The client receives the DHCPACK message with configuration
parameters. The client SHOULD perform a final check on the
parameters (e.g., ARP for
On Fri, 23 Nov 2007, Denys Vlasenko wrote:
+ length = sizeof(packet-options);
Yes. I'll do that on the magic numbers.
It's already done in svn. Take a look.
Brilliant ;)
I did look through and found a few more places.
The 576 is still magic. I think it would be better with
On Fri, 23 Nov 2007, Cristian Ionescu-Idbohrn wrote:
Still... This tweak is needed _only_ if the dhcp-server is brain
dead, problem which most clients don't suffer of.
This is what I could come out with. Please comment.
Cheers,
--
CristianIndex: networking/udhcp/Config.in
On Sat, 24 Nov 2007, Cristian Ionescu-Idbohrn wrote:
On Fri, 23 Nov 2007, Cristian Ionescu-Idbohrn wrote:
Still... This tweak is needed _only_ if the dhcp-server is brain
dead, problem which most clients don't suffer of.
This is what I could come out with. Please comment
On Sat, 24 Nov 2007, walter harms wrote:
a few comments on you patch:
if (end + string[OPT_LEN] + 2 + 1 = UDHCP_OPTIONS_BUFF_SIZE_MIN) {
bb_error_msg(option 0x%02x did not fit into the packet,
string[OPT_CODE]);
you know what packet so
Maybe create that directory?
http://busybox.net/downloads/fixes-1.8.2/
Cheers,
--
Cristian
___
busybox mailing list
busybox@busybox.net
http://busybox.net/cgi-bin/mailman/listinfo/busybox
On Sat, 24 Nov 2007, Denys Vlasenko wrote:
+#define UDHCP_UDP_PACKET_SIZE_MIN ((sizeof(struct udp_dhcp_packet)) - \
+ (CONFIG_UDHCP_OPTIONS_SLACK_FOR_BUGGY_SERVERS))
+
/* Let's see whether compiler understood us right */
struct BUG_bad_sizeof_struct_udp_dhcp_packet {
char
On Sun, 25 Nov 2007, Cristian Ionescu-Idbohrn wrote:
On Sat, 24 Nov 2007, Denys Vlasenko wrote:
Yes.
The check should have literal 576 - we are checking than neither we
nor compiler goof up.
That's fine.
But, please, use a '#define' for those 576 magic number.
Cheers,
--
Cristian
Denys,
Could you please comment on this:
On Fri, 23 Nov 2007, Cristian Ionescu-Idbohrn wrote:
Unless I'm missing something obvious, I have dificulties figuring out
how (on the same command line) to follow both rfc recommendations (the
60s and the 10s). If the dhcp-server doesn't answer, we
On Tue, 27 Nov 2007, Denys Vlasenko wrote:
Failed to reproduce it here.
Ok. I'll dig deeper.
Please send exact .config.
I'll do that too if I fail to find the problem.
Which versions of gcc
cris-axis-elf-gcc (GCC) 3.2.1 Axis release R64/1.64
ld
GNU ld version 2.12.1
Which libc
On Tue, 27 Nov 2007, Cristian Ionescu-Idbohrn wrote:
On Tue, 27 Nov 2007, Denys Vlasenko wrote:
Failed to reproduce it here.
Ok. I'll dig deeper.
Looks like its some sort of artifact of these 2 conf options.
CONFIG_FEATURE_VI_MAX_LEN=2048
CONFIG_FEATURE_VI_DOT_CMD=y
does not segfault
On Tue, 27 Nov 2007, Alexander Griesser wrote:
Sorry, no luck here in trying to reproduce this.
Thanks anyway.
Cheers,
--
Cristian
___
busybox mailing list
busybox@busybox.net
http://busybox.net/cgi-bin/mailman/listinfo/busybox
On Tue, 27 Nov 2007, Cristian Ionescu-Idbohrn wrote:
On Tue, 27 Nov 2007, Cristian Ionescu-Idbohrn wrote:
On Tue, 27 Nov 2007, Denys Vlasenko wrote:
Failed to reproduce it here.
Ok. I'll dig deeper.
Made some new discoveries. This is the latest result (see
attachment):
vi: HERE
I think I found the problem.
It seems to be caused by the poor correlation control between BUFSIZ,
COMMON_BUFSIZE and MAX_LINELEN and somewhat poor choice of macro
names. In this case:
, [ include/libbb.h ]
| 1101: #ifndef BUFSIZ
| 1102: #define BUFSIZ 4096
| 1103: #endif
`
BUFSIZ is
This is bb from svn HEAD.
On Sat, 8 Dec 2007, Denys Vlasenko wrote:
It's simpler than that. Programs should not make assumptions that
BUFSIZ is bigger than some fixed number. (Well, probably you can
be sure that it is bigger than, say, 63 :) ).
I configured bb with very small buffers:
As I did not get any reaction on this post:
http://busybox.net/lists/busybox/2007-December/029525.html
I'll start a new thread. Please refer to the link above.
Please find attached a minimal .config and a data file which can help
reproduce the problem. Edit that file and place the cursor on
On Fri, 21 Dec 2007, Denys Vlasenko wrote:
On Tuesday 18 December 2007 18:46, Cristian Ionescu-Idbohrn wrote:
As I did not get any reaction on this post:
http://busybox.net/lists/busybox/2007-December/029525.html
I'll start a new thread. Please refer to the link above.
Please find
Please take a look at this small patch. Should not change behaviour, just
make the code easier to read and avoid unnecessary repetitions.
Cheers,
--
CristianIndex: libbb/copyfd.c
===
--- libbb/copyfd.c (revision 20664)
+++
On Tue, 25 Dec 2007, Denys Vlasenko wrote:
On Tuesday 25 December 2007 12:08, Cristian Ionescu-Idbohrn wrote:
Seen a case when bb-netstat will output:
prompt ./busybox netstat -lanx
...
unix 2 [ ACC ] STREAM LISTENING 14220 private/mailman
unix 2
On Wed, 26 Dec 2007, Denys Vlasenko wrote:
On Tuesday 25 December 2007 16:34, Cristian Ionescu-Idbohrn wrote:
Yes. But it's too big. Let's see if attached few lines of
/proc/net/unix will suffice.
This line is especially ugly:
f7ab6280: 0002 0002 01 14954
On Thu, 27 Dec 2007, Cristian Ionescu-Idbohrn wrote:
I guess what I'm trying to say is that:
* if bogus lines detection is important, buffer sizes need to be used
* if bogus lines detection is important, buffer sizes and strlen() may
need to be used
* bogus lines should not be displayed
On Wed, 26 Dec 2007, Denys Vlasenko wrote:
On Wednesday 26 December 2007 20:42, Cristian Ionescu-Idbohrn wrote:
As log as _no attempt_ to display the bogus lines is made all should
be good. None of the other 3 warnings make any attempts to display
the bogus lines.
...leaving users
On Sun, 30 Dec 2007, Denys Vlasenko wrote:
Are you saying that this:
netstat: /proc/net/unix: bogus data on line 3
is *worse* than what you have shown above?
There must be some sort of confusion/misunderstanding somewhere :(
Let me try again. net-tools netstat shows this:
unix 2 [
r20695 | vda | 2007-12-30 02:59:53 +0100 (Sun, 30 Dec 2007) | 13 lines
libbb: introduce fputc_printable (from ed)
Prototype is here:
include/libbb.h:void fputc_printable(int ch, FILE *file);
Called from:
editors/ed.c: fputc_printable(ch | PRINTABLE_META, stdout);
On Sun, 30 Dec 2007, Cristian Ionescu-Idbohrn wrote:
There must be some sort of confusion/misunderstanding somewhere :(
Let me try again. net-tools netstat shows this:
unix 2 [ ACC ] STREAM LISTENING 14220private/mailman
unix 2 [ ] DGRAM
On Wed, 2 Jan 2008, Denys Vlasenko wrote:
The thread started by you reporting a bug where netstat crashed trying
to print binary data to stdout. This was fixed.
Yes, and I'm happy with that fix.
Now it even prints non-printable characters safely.
And that's very good too.
The only
On Sun, 6 Jan 2008, Denys Vlasenko wrote:
Thanks for the report. Using %-8.8s instead of %-8s fixes this.
I am also fixing VSZ overflow with large VSZs.
Will commit changes to svn in a minute.
Thanks.
The attached patch properly adjusts the columns to the header and
immitates debian ps
On Sun, 6 Jan 2008, Detlef Vollmann wrote:
+ smart_ulltoa4( (u, buf5, mgtpezy);
^^
This line contains a type, leading to:
procps/ps.c: In function `put_lu':
procps/ps.c:193: warning: left-hand operand of comma expression has no effect
procps/ps.c:193: warning:
On Mon, 7 Jan 2008, Denys Vlasenko wrote:
On Sunday 06 January 2008 14:16, Cristian Ionescu-Idbohrn wrote:
- len = printf(%5u %-8.8s %s %s ,
+ len = printf(%5u %-8.8s %6s %-4s ,
p-pid, user
On Tue, 8 Jan 2008, Cristian Ionescu-Idbohrn wrote:
The current svn chops of long usernames to 8 chars. That may create
confusion if the output from ps is in any way used to identify
process owners. Users:
'longnameA' and 'longnameB' will both be displayed as 'longname
On Wed, 9 Jan 2008, Lombard, David N wrote:
Just to be sure, this doesn't guarantee unique output. An integer
string is a perfectly valid username, both on a full distro and
bb/uc (just tested both). It may be quite demented, but this is
legal in /etc/passwd:
42:...:100:100:user
On Thu, 10 Jan 2008, Tito wrote:
Isn't there a maximum lenght for usernames on unix/linux?
Would it make sense to set the width of this field to that value?
Doing so seems like the clever thing to do, though you'd run into a
columnisation problem and loose compatibility to full version of the
On Thu, 10 Jan 2008, Tito wrote:
OTOH it doesn't solve the problem of a possible ambiguity
between username and uid as names composed only of digits
are legal.
What I meant was: either (default) truncated usernames _or_ special option
to show _only_ uids. That should surely solve the
The problem is described here too:
http://archives.devshed.com/forums/networking-100/dhclient-raw-socket-2147170.html
We noticed that too (corner cases). udhcpc eats up to 30% cpu.
That happens after DISCOVER while waiting for OFFER. udhcpc is looking
through every udp package.
I also
On Sun, 13 Jan 2008, Cristian Ionescu-Idbohrn wrote:
The problem is described here too:
http://archives.devshed.com/forums/networking-100/dhclient-raw-socket-2147170.html
We noticed that too (corner cases). udhcpc eats up to 30% cpu.
That happens after DISCOVER while waiting for OFFER
The '-l' option is documented but not supported:
, [ coreutils/stat.c ]
| getopt32(argv, ftL
| USE_SELINUX(Z)
| USE_FEATURE_STAT_FORMAT(c:, format)
| );
`
, [ include/usage.h ]
| #define stat_full_usage \
|
On Thu, 17 Jan 2008, Paul Fox wrote:
+ if (setsockopt(fd, SOL_SOCKET, SO_ATTACH_FILTER, filter_prog,
+ sizeof(filter_prog)) 0)
+ bb_perror_msg_and_die(setsockopt);
+ DEBUG(attached filter to raw socket fdr %d, fd);
is this supported on all kernel
Oh, Bernhard...
Why do you make my life so much more complicated than it already is ;)
On Thu, 17 Jan 2008, Bernhard Fischer wrote:
Can you please separate the whitespace fixes from actual changes
(diff -rdupbBw or the like to filter out the noise).
Wasn' much of that. Attached is the
On Thu, 17 Jan 2008, Stefan Rompf wrote:
Am Mittwoch, 16. Januar 2008 23:28 schrieb Cristian Ionescu-Idbohrn:
It's perfectly fine with me if you use this code in udhcpc, also
with later versions of the GPL as your license allows.
Great. Thanks.
One minor nitpick:
+ if (setsockopt
On Fri, 18 Jan 2008, Stefan Rompf wrote:
Am Donnerstag, 17. Januar 2008 16:49 schrieb Cristian Ionescu-Idbohrn:
(void)setsockopt(sock, SOL_SOCKET, SO_ATTACH_FILTER, filter,
sizeof(filter));
No error checking at all.
Feature ;-)
I thought you'd say that ;)
I knew some failures
On Wed, 23 Jan 2008, Stefan Rompf wrote:
Am Freitag, 18. Januar 2008 21:54 schrieb Cristian Ionescu-Idbohrn:
Makes me happy too ;) But the credit should go to a collegue of
mine (Bcc:ed) who checked the code and suggested that.
I can credit to unknown colleague of Cristian Ionescu
On Sun, 13 Jan 2008, Cristian Ionescu-Idbohrn wrote:
The problem is described here too:
http://archives.devshed.com/forums/networking-100/dhclient-raw-socket-2147170.html
We noticed that too (corner cases). udhcpc eats up to 30% cpu.
That happens after DISCOVER while waiting for OFFER
On Fri, 25 Jan 2008, Denis Vlasenko wrote:
On Friday 25 January 2008 14:02, Cristian Ionescu-Idbohrn wrote:
Is there anything I can do to attract more attention to the patch I
sent which cures the problem mentioned above?
Sorry, I was (and still is) busy with job search.
No worries
CC networking/route.o
make[1]: *** No rule to make target `networking/sendmail.o', needed by
`networking/lib.a'. Stop.
Cheers,
--
Cristian
___
busybox mailing list
busybox@busybox.net
http://busybox.net/cgi-bin/mailman/listinfo/busybox
On Wed, 30 Jan 2008, Cristian Ionescu-Idbohrn wrote:
Looks like last checkin with this comment:
,
| r20923 | aldot | 2008-01-29 11:33:34 +0100 (Tue, 29 Jan 2008) | 4 lines
| - be C99 friendly. Anonymous unions are a GNU extension. This change is
| size-neutral WRT -std=gnu99 and fixes
Would this adjustment be appropriate?
No point in calling get_cached_username twice?
Index: procps/ps.c
===
--- procps/ps.c (revision 20936)
+++ procps/ps.c (working copy)
@@ -4,7 +4,7 @@
*
* Copyright (C) 1999-2004 by Erik
On Wed, 6 Feb 2008, Steven Shiau wrote:
I can successfully compile this svn ver 20945 on Debian etch and Ubuntu
7.10.
Yes, later vintage distribution.
However, it failed in Redhat 9 (gcc-3.2.2-5, kernel 2.4.33.4-6):
Isn't that an older/unsupported/patched in incompatible ways dist? Oh,
Fixes warning about _GNU_SOURCE being redefined.
Typo in comment.
Cheers,
--
CristianIndex: shell/ash.c
===
--- shell/ash.c (revision 20984)
+++ shell/ash.c (working copy)
@@ -51,8 +51,10 @@
#endif
#if DEBUG
+#ifndef
I'd like to decouple 'mount -f' from USE_FEATURE_MTAB_SUPPORT, unless
there are serious reasons against it.
mount man-page:
-f Causes everything to be done except for the actual system call;
if it's not obvious, this ``fakes'' mounting the file system.
This option is
On Sat, 16 Feb 2008, Cristian Ionescu-Idbohrn wrote:
On Thu, 14 Feb 2008, Cristian Ionescu-Idbohrn wrote:
I'd like to decouple 'mount -f' from USE_FEATURE_MTAB_SUPPORT, unless
there are serious reasons against it.
mount man-page:
-f Causes everything to be done except
On Thu, 14 Feb 2008, Cristian Ionescu-Idbohrn wrote:
I'd like to decouple 'mount -f' from USE_FEATURE_MTAB_SUPPORT, unless
there are serious reasons against it.
mount man-page:
-f Causes everything to be done except for the actual system call;
if it's not obvious
Trivial.
Cheers,
--
Cristian--- miscutils/Config.in 2008-02-19 08:30:28.0 +0100
+++ miscutils/Config.in 2008-02-19 11:05:10.0 +0100
@@ -27,6 +27,7 @@
config FEATURE_CHAT_NOFAIL
bool Enable NOFAIL expect strings
+ depends on CHAT
default y
help
When enabled expect
Cleanup whitespace damage.
Cheers,
--
Cristian--- util-linux/findfs.c 2008-02-19 08:30:28.0 +0100
+++ util-linux/findfs.c 2008-02-19 11:36:55.0 +0100
@@ -17,7 +17,7 @@
char *tmp = NULL;
if (argc != 2)
- bb_show_usage();
Trivial fix for build error when ENABLE_FEATURE_MOUNT_LABEL is
undefined.
Cheers,
--
Cristian--- util-linux/mount.c 2008-02-19 08:30:28.0 +0100
+++ util-linux/mount.c 2008-02-19 11:40:14.0 +0100
@@ -248,6 +248,7 @@
#define verbose_mount(...) mount(__VA_ARGS__)
#endif
+#if
r21020 is ok.
ash r21030 breabs in an init script:
rc (pid 75) segfaults for page address 8feee000 at pc 355a06a6
while starting a 'for' loop, I'm not sure, I'm investigating.
I'll let you know after digging some more.
Cheers,
--
Cristian
___
On Tue, 19 Feb 2008, Cristian Ionescu-Idbohrn wrote:
r21020 is ok.
ash r21030 breabs in an init script:
rc (pid 75) segfaults for page address 8feee000 at pc 355a06a6
while starting a 'for' loop, I'm not sure, I'm investigating.
I'll let you know after digging some more.
Yes
On Tue, 19 Feb 2008, Denys Vlasenko wrote:
Here:
/* A=a B=$A case: var_str_list is a list of A=a strings
* which should be considered before we check variables. */
if (var_str_list) {
Does it work again yf you replace if (var_str_list) with if
On Tue, 19 Feb 2008, Denys Vlasenko wrote:
On Tuesday 19 February 2008 22:12, Cristian Ionescu-Idbohrn wrote:
On Tue, 19 Feb 2008, Denys Vlasenko wrote:
Is this on x86 with .config I sent?
No. It's my config. Comes in an attachment.
I can reproduce it.
Alright.
But when I flip
On Wed, 20 Feb 2008, Denys Vlasenko wrote:
Yes. We were using uninitialized data on stack.
Right.
Fix is attached. Please test.
Looks much better.
I'll test on my embedded systems too.
Cheers,
--
Cristian
___
busybox mailing list
On Wed, 20 Feb 2008, Marc Leeman wrote:
And if you try to scp for the second time - eg. when the fingerprint
is already saved on your host? I've experienced similar problems with
earlier versions of busybox and it helped if fingerprint was already
known.
The fingerprint is never known:
Hi Denys,
On Wed, 20 Feb 2008, Cristian Ionescu-Idbohrn wrote:
On Wed, 20 Feb 2008, Denys Vlasenko wrote:
On Wednesday 20 February 2008 17:37, Denys Vlasenko wrote:
On Wednesday 20 February 2008 11:46, Cristian Ionescu-Idbohrn wrote:
I'm still seeing uninitialized data which leeds
GNU C version 3.3.6 (Debian 1:3.3.6-15) (i486-linux-gnu)
says:
networking/brctl.c:39: error: duplicate `inline'
networking/brctl.c:52: error: duplicate `inline'
line:
static inline ALWAYS_INLINE void strtotimeval(struct timeval *tv,
is expanded to:
static __inline__ __attribute__
Something like this seems to be required.
Cheers,
--
Cristian--- util-linux/mount.c.~3~ 2008-02-28 08:30:12.0 +0100
+++ util-linux/mount.c 2008-03-11 10:14:01.0 +0100
@@ -22,8 +22,10 @@
#include syslog.h
#include libbb.h
+#if ENABLE_FEATURE_MOUNT_LABEL
/* For
On Mon, 10 Mar 2008, Cristian Ionescu-Idbohrn wrote:
Alright. Back from holidays. Picked it up from where I left.
I first applied the patch. Still segfaults :(
Applied the 2 suggested 'memset' changes mentioned above. Kernel
boots, linuxrc starts running but ends quite fast showing
On Tue, 18 Mar 2008, Cristian Ionescu-Idbohrn wrote:
Please consider this test script:
I seem to get the behaviour I expect using the attached script, after
apllying the attached patch.
Cheers,
--
CristianIndex: procps/top.c
Please consider this test script:
---
#!/bin/bash
n=5
cmd=top -b -d 1 -n $n
(date +%T;echo $cmd;$cmd /dev/null; date +%T)
wait
echo
cmd=./busybox top -b -d 1 -n $n
(date +%T;echo $cmd;$cmd /dev/null; date +%T)
wait
---
Produces following results:
13:37:31
top -b -d 1 -n 5
13:37:35
On Mon, 24 Mar 2008, Paul Fox wrote:
denys wrote:
Trivial bashism example: function keyword.
exactly. trivial. delete it. :-)
Or just put it under a new option:
I_KNOW_AND_I_DONT_CARE_CAUSE_I_LIKE_BASH_BETTER
Bashisms in ash can be guarded by ENABLE_FEATURE_BASH_COMPAT,
in order to
The line above says it all.
Cheers,
--
Cristian--- util-linux/mount.c (revision 21509)
+++ util-linux/mount.c (working copy)
@@ -23,7 +23,6 @@
#include libbb.h
#if ENABLE_FEATURE_MOUNT_LABEL
-/* For FEATURE_MOUNT_LABEL only */
#include volume_id.h
#endif
This one: ${parameter#word}. Can be reproduced with:
# ./ash -c 'var=1/2/3;echo ${var#?}
1/2/3
expected: /2/3.'
Cheers,
--
Cristian
___
busybox mailing list
busybox@busybox.net
http://busybox.net/cgi-bin/mailman/listinfo/busybox
On Fri, 11 Apr 2008, James Simmons wrote:
On Fri, 11 Apr 2008, James Simmons wrote:
This one: ${parameter#word}. Can be reproduced with:
# ./ash -c 'var=1/2/3;echo ${var#?}
1/2/3
expected: /2/3.'
Did this work before?
But of course:
# busybox
Cheers,
--
CristianIndex: networking/ipcalc.c
===
--- networking/ipcalc.c (revision 21711)
+++ networking/ipcalc.c (working copy)
@@ -181,7 +181,7 @@
bb_herror_msg_and_die(cannot find hostname for %s, argv[0]);
}
Please consider the example code in the attachment. That script is
supposed to show that a simple 'tr'-like shell function is faster than a
real forked 'tr'.
My empirical numbers show a factor 5 better performance for the shell_tr
function (ash and dash). Just for amusement I even tested the
There are some performance down sides if one makes intensive use of
certain bb-applets in shell scripts. And that's the probable reason why
2 of these applets (echo and test) are available as BUILTINs.
Which of these simplier applets could become builtin candidates which are
easy enough to add
On Sat, 12 Apr 2008, Cristian Ionescu-Idbohrn wrote:
On Sat, 12 Apr 2008, Denys Vlasenko wrote:
This patch should make it even faster by not forking for all NOFORK
applets.
Can you give it a try?
But of course. Please give me until tomorrow.
Looks good for most of the applets I
On Sun, 13 Apr 2008, Denys Vlasenko wrote:
From your list above, kill is already a builtin,
Hmm... I don't see it. include/applets.h says:
USE_KILL(APPLET(kill, _BB_DIR_BIN, _BB_SUID_NEVER))
meaning kill isn't a NOFORK/NOEXEC applet?
Yes, it isn't. It can probably made to be
On Sun, 13 Apr 2008, Denys Vlasenko wrote:
I remember now what gives the most headache with NOFORKs in shells.
It's Ctrl-Z.
What should happen if you run sleep 10 in interactive, NOFORK-enabled
shell, and then hit Ctrl-Z? There is _no process_ to_ background_!
True. Anything builtin can't
Wrote this small script (see attachment) to help me test ash's performance
(fork vs. nofork/noexec). Run like this:
, [ builtin rm ]
| # ash-loop-cmd.sh 1 'touch /tmp/x;rm -f /tmp/x'
| real0m1.179s
| user0m0.268s
| sys 0m0.800s
`
, [ forked rm ]
| # ash-loop-cmd.sh
On Sun, 13 Apr 2008, Denys Vlasenko wrote:
tac cannot be NOFORK as written because it uses xrealloc:
I should probably read the code before asking stupid questions.
In order to make tac NOFORK, it needs to be rewritten to cleanly
free all allocated memory prior to exit.
Of course.
Cheers,
On Tue, 15 Apr 2008, James Simmons wrote:
This fails.
runlevel=$(cat /proc/1/cmdline 2 /dev/null)
echo $runlevel# init [2]
runlevel=${runlevel#* [}
echo runlevel '$runlevel' # Should be runlevel '2]'
On Wed, 16 Apr 2008, Paul Smith wrote:
On Wed, 2008-04-16 at 17:59 +0200, Cristian Ionescu-Idbohrn wrote:
Please consider the 1st attachment as a POC, and then the 2nd
attachment, which is more dearer to me ;-)
???
+#define FIVETWELVE 512
How is using FIVETWELVE any better than 512
On Wed, 16 Apr 2008, Denys Vlasenko wrote:
On Wednesday 16 April 2008 18:33, Paul Smith wrote:
On Wed, 2008-04-16 at 17:59 +0200, Cristian Ionescu-Idbohrn wrote:
Please consider the 1st attachment as a POC, and then the 2nd
attachment, which is more dearer to me ;-)
???
+#define
1 - 100 of 492 matches
Mail list logo
