I have built Busybox v1.19.4 with Buildroot for my target device and added
the sendmail command.
The problem is that I am not allowed to tell sendmail to stop getting data
from stdin and finally send the email.
I have tried in the following ways:
1. sendmail -f f...@domain.com -S
Wonder whether it's feasible if I provide a patch for busybox to take
and ignore --no-cache long option?
Yes please.
Please, consider applying.
TIA,
--Vladimir
wget-no-cache.diff
Description: Binary data
___
busybox mailing list
Hi!
Recent luarocks (package manager for Lua) use `wget --no-cache ...` to
fetch data.
Wonder whether it's feasible if I provide a patch for busybox to take
and ignore --no-cache long option?
TIA,
--Vladimir
___
busybox mailing list
busybox@busybox.net
/etc/mdev.conf
I guess we have this one under ./examples. Please, check it
--Vladimir
___
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
Thanks Vladimir, I did already, but it doesn't go into the use of modprobe.
check L16 of ./examples/mdev_fat.conf -- mdev does honor $MODALIAS
coming from the kernel uevents
--Vladimir
___
busybox mailing list
busybox@busybox.net
the hardware (the network cards in my case) for which the appropriate
module is not loaded yet. It doesn't scan the all PCI devices
Right. mdev is for hotplugging. My patch for coldplugging (and imho
more accurate hotplugging) branch was NAKed. You may try it:
scripts/basic/split-include.c: In function ‘main’:
scripts/basic/split-include.c:134: warning: ignoring return value of
‘fgets’, declared with attribute warn_unused_result
scripts/kconfig/conf.c: In function ‘conf_askvalue’:
scripts/kconfig/conf.c:104: warning: ignoring return value of
Thanks for the script Andrew! I'll incorporate it (and possibly use it
permenantly) right away. Also, I am currently using the full version of
modutils offered by busybox, so no patching for me! :)
This is the best.
___
busybox mailing list
Thanks for the script Andrew! I'll incorporate it (and possibly use it
permenantly) right away. Also, I am currently using the full version of
modutils offered by busybox, so no patching for me! :)
BB modutils is quite good. Stay with them. Use Andrew's script for
coldplugging and BB mdev
Thanks a lot!
On Thu, Jan 6, 2011 at 1:22 PM, Rob Landley r...@landley.net wrote:
On Thursday 06 January 2011 04:07:39 Peter Korsgaard wrote:
Rob == Rob Landley r...@landley.net writes:
Rob The solution embedded people use is either:
Rob 1) Use a jtag to reflash the device, so no matter
Hi!
We at #openlgtv have been using flash_eraseall /dev/mtdX and then
cat newfile /dev/mtdX. Since there's no atomicity in such a
process, chances high to brick the device.
Wonder how safe is to use BB flashcp utility?
Best regards and TIA,
--Vladimir
P.S.
Posted to the author (To:
-m [ type ] - create a multipart mime section from file of this
Content-Type: (default is multipart/mixed).
-j file1 file2 - join mime section file2 to multipart section file1.
Can you _try to mimick_ -m and -j by means of a shell script? I recall
I express didn't code
... by sending e-mails directly using busybox with a small shell script...
That's it. What does this script do? Mind to share?
___
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
The latest sendmail command syntax is a bit difficult to work out from
the help text ( for a novice )
We tried to make BB sendmail just compliant, so any `man sendmail` should help
Also Read additional recipients from message body doesn't actually tell
you what format to put them in, or
When I was doing BB sendmail, I was stumbled upon I had no free
letters in ABC to denote options they invented :)
___
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
Hi!
Attached is the patch to fix naive assumption that critical headers
should have at least one space after the colon.
Denys, please, consider applying
TIA,
--Vladimir
sendmail.patch
Description: Binary data
___
busybox mailing list
I'd propose to consider mangling make script making BB coreutils as
0-step and then use them to build the target real BB. 0-step BB
should be compiled w/o help messages and depend on the stuff truly
expected to be core. That list to be defined.
--Vladimir
... if I pipe a Mail formatted with makemime to reformime it aborts with
a read error. The file gets created on output, but is always empty. I do
currently not have any other mime formatted messages on that machine to
test, but it looks like the bug is in reformime. The error happens on
pipe
Please, run
$ strace -f -v -s1024 -o log.txt reformime zzz
and mail here log.txt attached
--Vladimir
___
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
Sorry, no strace on that machine installed. Doesn't the error happen on
your system?
I've no BB setup within my grasp.
Can someone on the list help us?
TIA,
--Vladimir
___
busybox mailing list
busybox@busybox.net
While we are at it, why mailutils contain almost exact copy of uudecode's
base64 encoder and decoder?
Lemme recall. Can you point out the diffs you noticed?
--Vladimir
___
busybox mailing list
busybox@busybox.net
Why don't you create a single symlink
pointing to the absolute path of busybox and do hardlinks to the symlink
for the remaining applets?
E.g. /sbin, /bin, /usr/sbin, /usr/bin can be on various filesystems,
thus hardlinking will fail
--Vladimir
___
Lemme recall. Can you point out the diffs you noticed?
I already merged them.
http://git.busybox.net/busybox/commit/?id=9fe98f701d40835db32baa12c94b661d40231ea4
mail's one wanted '-' as a terminator. uudecode wanted
base64 wanted EOF.
Oh. I see. I guess I used a clone because of it
I agree. Someone needs to create a patch.
Definitely. Pity, I quite deviated from C development in favour on JS
[off]node.js is my hottest area[/off].
Will a cmdline switch suffice which selects line endings?
--Vladimir
___
busybox mailing list
Better to always accept both \n and \r\n as valid
How to distinguish \r being a part _of the real data_ from \r being a
part of CRNL then? Heuristics may lead to unwanted data mangling...
--Vladimir
___
busybox mailing list
busybox@busybox.net
In that case you just need a single symlink on each filesystem ... and
how much systems do you know those directories are on different
filesystems today?
Mine was and I hope still is, at least. And probably every except yours?
It's a matter of BB installation script, anyway
--Vladimir
just drop (ignore) \r and only count \n as line termination. The better
way is to check, if \r is followed by \n otherwise it is a formatting
error.
I won't step inside the discussion on a very loosely defined terms.
The essense:
--
Right now \r\n\r\n is hardcoded as EOL. This allows for
In the good old days, /usr may had gone to a separate _disk_, ...
Folks. Frankly, you say disks _are_ cheap and big nowadays, and at the
same time you keep worrying about 200 inodes. Let really little-spaced
systems be free to use more tricky installations, including hardlink
technique, on
No reading ahead! Without breaking anything. You see?
I see. Have you looked at the code? I tried my best in optimizing the
parser. That is why xmalloc_fgets_str(). Just take a look at the code
and the comments.
--Vladimir
___
busybox mailing list
Have you looked at the code?
Busybox source is currently on another machine, so it is a bit difficult
to look in.
http://git.busybox.net/busybox/tree/mailutils/mime.c
Seems you are doing a line buffer approach. I'm not familar
with libbb functions, but ther seems to be a function
In upstream that works fine, but busybox mount can't find the mount point
in fstab if you specify only a relative path (not starting with a slash)
and the current working directory is not the root directory.
Personally, I'd say it's mainstream mount who looks weird.
--Vladimir
Problem with filesystem mount happens again.
Just to be sure --- you have 4 primary partitions, right?
--Vladimir
___
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
* You do not dispute usefulness of that feature, but feel that
another feature, namely creating gzipped pages on demand,
will be also useful
Folks. Consider another subtle issues with the current approach:
* if you GET /Foo.gz , httpd is supposed to regzip it, since vanilla
browser
Folks. Consider another subtle issues with the current approach:
Sorry for buzz, irrelevant.
--Vladimir
___
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
Vladimir Consider modern nginx (and I assume apache also) setups which still
Vladimir inhibits gzipping for IE 7.
http://www.contentwithstyle.co.uk/content/moddeflate-and-ie6-bug
http://robsanheim.com/2008/02/07/beware-the-default-nginx-config-old-ie6-hates-gzip/
But I admit we can just
Accept-Encoding:
Just FYI, Internet Explorer 7 sends Accept-Encoding: gzip but cannot
correctly process gzipped response. You should analyze also
User-Agent: and conditionally skip gzipping.
Best regards,
--Vladimir
___
busybox mailing list
Vladimir Just FYI, Internet Explorer 7 sends Accept-Encoding: gzip
Vladimir but cannot correctly process gzipped response. You should
Vladimir analyze also User-Agent: and conditionally skip gzipping.
Really? I have an IE6 running here in wine, and it seems to work ok:
Consider modern
I decided to stop doing it myself (poorly), and start using
Rob's Aboriginal Linux:
http://impactlinux.com/aboriginal/
Great news!
--Vladimir
___
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
in fact the problem was not is what mdev was doing but that it was not
getting triggered at all.
You welcome. Enjoy busybox ;)
--Vladimir
___
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
inotifyd.c:(.text.inotifyd_main+0x20): undefined reference to
`inotify_init'
Kernel version?
-- Vladimir
___
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
make CROSS_COMPILE=/usr/local/arc.4.2.1/sk885x-2.6/arc-linux-uclibc/bin/
Again. To what _target_ kernel version your cross toolchain is bound?
Find sys/inotify.h there.
-- Vladimir
___
busybox mailing list
busybox@busybox.net
and yes, I do find the notify.h in include/linux/ in kernel source tree.
What does intotifyd.c include, related to inotify.h?
Try to replace
#include sys/inotify.h
with
#include /absolute/path/to/sys/inotify/in/your/toolchain.h
If not, try to replace
#include sys/inotify.h
with
#include
Hi!
Does anybody try to push to BB ftpd file/dir containing non-latin1
letters in the name?
IOW, is the subj true?
TIA,
-- Vladimir
___
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
# popmaildir -k /var/spool/mail -- nc pop.gobizmail.com 110
User:
Password:
popmaildir: can't open 'tmp/94668534938.1114.(none)': No such file
or directory
popmaildir: helper killed by signal 15
#
Here, I enter the username and password at the prompt and then throws
the error
Thanks a ton for continued help .
That works.
Welcome!
BTW, this is the cpu info:
# cat /proc/cpuinfo
Processor Number : 0
Processor version : 0 (ARCtangent A4 processor family)
CPU speed : 162.00 Mhz
Bogo MIPS : 80.89
Timers: TIMER1 TIMER0
INT_VECBASE:
I am trying to start popmaildir with :
popmaildir -k /var/spool/mail pop.gobizmail.com
where I am expecting the mail at /var/spool/mail and my pop server is
pop.gobizmail.com ( i use the same settings on outlook too).
I ask you _again_, what are these settings on outlook too?
In deep,
I checked the netcat behaviour for my pop server, this is what I get:
# nc pop.gobizmail.com 110
+OK POP3 TIMS(5.40.2008012515) server ready.
14451.1268202...@sme.or.kr
OK. Try to issue USER your-username. What is the answer?
is this correct .. sorry, I don't have the strace yet .
By
On Saturday 09 January 2010 22:53:41 Zvi Vered wrote:
in /share/terminfo/l there is a file: linux
You can also try to place this file under:
/usr/share/terminfo/l/
/usr/lib/terminfo/l/
Or, better, run
# strace -f -v -s1024 -o log htop
then check log file for paths containing terminfo and
/usr/sbin/sendmail: invalid option -- i
but to answer your question, seems like git-send-email does something
like: /usr/sbin/sendmail -i -ffromaddress
sendmail.c:
opts = getopt32(argv, tf:o:iw:H:S:a::, opt_from, NULL, timeout,
opt_connect, opt_connect, list);
AFAIKS, sendmail -i
no, it does not segfault. something is calling tune2fs directly after
mounting the rootfs.
We have to determine that something. Drop here your boot scripts.
--
Vladimir
___
busybox mailing list
busybox@busybox.net
i deleted 'tune2fs' and now system boots fine. but there were no messages why
tune2fs killed init.
OK. Don't delete tune2fs but rename it to tune2fs.0. Boot your system.
Rename tune2fs.0 back. Try it. Does it segfault?
--
Vladimir
___
busybox mailing
Why rd/*? why loop/*? Is it required for all Linux systems
I dont remember. It was something I found that gentoo udev did and
copied it. Might be that it was something udev used to do but no longer
does.
Looks like those are not needed.
It was to mimic devfs-like organization of /dev, to
Line 87 of env.c should be
for (ep = environ; ep *ep; ep++) {
instead of
for (ep = environ; *ep; ep++) {
or plain env - segfaults.
Also, wonder how one can clear current environment within a shell script?
--
Vladimir
___
busybox
Using export -n or unset?
I need command to effectively perform clearenv(), to unset all vars at
once ( that's why I tried env - :). Is there such in the nature?
--
Vladimir
___
busybox mailing list
busybox@busybox.net
* ndev -S to replay all missed hotplug events -- you case, afaics
* ndev during runtime to process ordinary hotplug events
Try ndev -S
--
Vladimir
Regards
~Sameer
On Thu, Nov 12, 2009 at 12:54 AM, Vladimir Dronnikov
dronni...@gmail.com wrote:
:)
vanilla i get.
you are refering to vanilla
However, i have a similar _hotplug script to create symbolic links to
event* devices in /dev/input/ .This does not work is the system is
booted with the input device connected, whether or not evdev is compiled
as a static or dynamically loadable module
hotplug events for plugged-in devices
Can't we apply proposed patches, those being guarded by, say,
FEATURE_ASH_SANE_TTY. That way the code won't go in vain plus some
users won't be forced to keep their private patchsets.
Regards,
--
Vladimir
___
busybox mailing list
busybox@busybox.net
I tried to write a version of lspci, it only uses sysfs and supports 2
options: m and k, it dumps data only in numeric form (not uses pci.ids file)
Looks familiar :)
BTW, updated ndev is at http://busybox.net/~dvv/patch/ndev.patch :
speedup; diving into /sys/devices is moved to a separate
would probably be good anyways as i imagine a sysfs-based lsusb would utilize
the code too
Right. This would obsolete two external dependencies of my system.
What would be the correct dancing about clearenv()/setenv()/putenv()
within a potentially long cycle?
--
Vladimir
# null may already exist; therefore ownership has to be changed with command
null root:root 666 @chmod 666 $MDEV
In fact, this is funnier than the code funny snippet we discussed
earlier (about assert) -- this one reads: make it 666, then make it
666 again.
I think the node should
* http://busybox.net/~dvv/patch/hotplug.patch : makes mdev available
under hotplug alias -- this allows it for seamless usage as hotplug
helper, no need to remember to pass /sbin/mdev to
/proc/sys/kernel/hotplug
Not sure this is useful enough, let's put it on hold for now?
Sure.
--
vda
While I'm at it, would be nice with an /etc/mdev.d/ dir where packages
like qemu could install an /etc/mdev.d/kvm file and similar.
sounds like a reasonable config option and should be easy to support. patches
welcome ;).
Put in mind ability to use envdir. In essence, our mdev.conf can be
No. It may have sense sometime ago, when chmod was skipped
if /dev/null existed, but command was executed even if it
did exist. Maybe it was working this way many versions ago.
This or that, fun is that mdev users/developers had (still have?) to
workaround their own tool.
BTW, can't help
Hello!
* http://busybox.net/~dvv/patch/swap.patch : libvolume_id is now
capable of recognizing swap partitions which contain suspended system
image. This is useful upon resume from hibernation, when the correct
resume=major:minor cmdline kernel parameter is to be supplied. Now one
can use findfs
a
function that resolves mount specs to device names.
However, we do not want to diverge much from standard utilities. Can
you add the feature to standard mountpoint?
Best regards,
--
Vladimir Dronnikov, a busybox developer, Russia
___
busybox mailing list
How does the big mountpoint handle UUIDs, if at all? If it does not, can
you suggest such a feature to it's upstream so we do not diverge?
First of all, I think nobody forbids us to be more functional than
upstream original. E.g., BB mount is very convenient with its
automatic -o loop. And
runsv spoils service container directory with supervise directory --
pure runtime info which _should not_ survive between reboots -- thus
it's native place is /var/run/service name.
There too should go supervise or service logger subservice.
If you like stuff to be spread into many
Hello, Denys!
I'd like to discuss the drawback (feature?) of runit services
(http://busybox.net/~vda/init_vs_runsv.html) which has always been
inspiring me to make some changes:
runsv spoils service container directory with supervise directory --
pure runtime info which _should not_ survive
util-linux/volume_id/btrfs.c:32: error: expected specifier-qualifier-list
before '__le64'
That's why using uintNN_t is preferred. Those are standard types.
They will always work.
I see.
I wonder why kernel uses __u32 etc instead, with no sign
of gradual migration to uint32_t?
(I
Hi!
Two tiny patches:
* http://busybox.net/~dvv/patch/fbsplash.patch : makes fbsplash able
to handle zipped images, when image is specified via -s option
* http://busybox.net/~dvv/patch/hotplug.patch : makes mdev available
under hotplug alias -- this allows it for seamless usage as hotplug
if (mknod(node_name, mode | type,
makedev(major, minor)) errno != EEXIST)
bb_perror_msg_and_die(can't create
%s, node_name);
It exits if mknod fails with errno != EEXIST.
This means that /dev itself does not exist,
is
ELOOP
Too many symbolic links were encountered in resolving pathname.
mdev only creates /dev/foo
Wrong. /dev/what/ever/you/link/foo is ok. See build_alias(), which
calls bb_make_directory().
ENAMETOOLONG
pathname was too long.
i guess if /dev was a symlink to
util-linux/volume_id/btrfs.c:32: error: expected specifier-qualifier-list
before '__le64'
util-linux/volume_id/btrfs.c:51: error: expected specifier-qualifier-list
before '__le64'
util-linux/volume_id/btrfs.c: In function 'volume_id_probe_btrfs':
util-linux/volume_id/btrfs.c:92: error:
Hello!
line 334 of mdev.c contains:
bb_perror_msg_and_die(can't create %s, node_name);
IMHO, it should be just bb_perror_msg(), or we may have some devices missed.
Regards,
--
Vladimir
___
busybox mailing list
busybox@busybox.net
Hello, list!
I've been developing my flavor of BB mdev, but eventually realized it
takes much effort to patch it. As the patch itself grew more than
original utility code, I decided to leave mdev alone and to move
changes to a new applet -- ndev -- source attached.
Reasons to do that:
* I
Hi, Michele!
Since .ppm files are rather big, a zipped ones would be nice to have.
Attached is the patch for fbsplash to cope with compressed image files
via standard open_zipped() procedure.
Of course we could pipe image, say: gunzip fbsplash -s -, but this
reserves stdin for image, and we
cd /sysroot
pivot_root . old_root
umount -n /old_root
exec chroot . /sbin/init
First try without umount -n /old_root. When after /sbin/init is run
you are in console, try umount /old_root manually. Does it work?
BTW, what is -n option for umount?
--
Vladimir
1. what does mount say? /old_root is mounted?
2. try umount -l /old_root. Does it work?
--
Vladimir
___
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
Hello, Bernhard!
mountpoint -n PATH refuses to find out the name of mounted device in
case of so-called anonymous superblock, e.g. for btrfs.
That is, reported st.st_rdev doesn't correspond to any block device under /dev
http://www.mail-archive.com/linux-bt...@vger.kernel.org/msg02870.html
To
Hello!
Attached is the fix to enable BB be built with WERROR defined unless
FEATURE_ASSUME_UNICODE or FEATURE_EDITING_ASK_TERMINAL are enabled.
Denys, please, consider applying.
TIA,
--
Vladimir
test.patch
Description: Binary data
___
busybox
Attached is the fix to enable BB be built with WERROR defined unless
FEATURE_ASSUME_UNICODE or FEATURE_EDITING_ASK_TERMINAL are enabled.
Sorry, the patch would break the logic -- discard it.
Denys, please try to build with my config -- unguarded unused label
emits compilation error.
--
Hello!
Attached is the patch for volume_id subsystem to recognize btrfs partitions.
Denys, please, consider applying.
TIA,
--
Vladimir
test.patch
Description: Binary data
___
busybox mailing list
busybox@busybox.net
i did not try; but what will happen if argc==1 ?
argv[1] should be undefined.
argv[1] then will be NULL, so the code is OK
--
Vladimir
___
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
The point of malloc() is to handle compressed modules. Do you use
compressed (.ko.gz, .ko.bz2) modules?
should be trivial to detect the compressed header and just use mmap() if it's
uncompressed
Of course. Still the question is whether mmap() consumes no memory :)
--
Vladimir
Hello, Clemens!
... I have tried busybox .14.x on an asus router (mips target) ...
Can you play with BB mke2fs on that target? IIRC, mips is big-endian.
TIA,
--
Vladimir
___
busybox mailing list
busybox@busybox.net
Hi, list!
I used to lay upon ls /some/dir dumps _just_ the content of
/some/dir. Now it prepends the output with total line _by
default_, though AFAIK shouldn't.
Is such a behavior right?
To track the issue: the line
---
* 1. ls -l of a directory doesn't give total blocks header
---
exists
In my testing, bare ls does not print total, ls -l does.
Please post your .config
--
Here it goes.
--
Vladimir
.config
Description: Binary data
___
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
this behavier is correct e.g. gnu-ls says:
ls -l /proc/
total 0
dr-xr-xr-x 7 root root 0 10. Okt 15:50 1
dr-xr-xr-x 7 root root 0 10. Okt 15:50 10
use sed 1d to get rid of it.
Notice, you used -l switch, Walter. What gnu-ls says on ls /proc/?
In my testing, bare ls does not print total, ls -l does.
The quirk is that bare ls /some/dir works, but redirected one ls
/some/dir 1 dumps total xxx!
--
Vladimir
___
busybox mailing list
busybox@busybox.net
ls /proc/
1 17456 2754 3476 3941 69 interrupts schedstat
no totals
Try
ls /proc 1; cat 1
--
Vladimir
___
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
Speaking of ftp.. Does anybody else find it bloated to have at least two
FTP talkers, one in wget and another one in ftpgetput? ;)
My situation: a task to periodically fetch the content of M$Exchange
mailbox via HTTP. BB wget can't be used -- it has no out-of-box
support for cookies. I invented
Insmod fails on system with small amout of memory. I traced this problem to
a change where insmod used malloc instead of mmemap. I wonder if we can add
a option which will select either of the methods?
The point of malloc() is to handle compressed modules. Do you use
compressed (.ko.gz,
With removed struct ext2_inode_large declaration and with added
sanity check on -I NUM, this patch will be ok.
Please, try attached patch.
--
Vladimir
inodesize.patch
Description: Binary data
___
busybox mailing list
busybox@busybox.net
Applied, thanks.
Shouldn't we finally try to attack -O ^resize_inode incompat? The
code (ext2fs_create_resize_inode()) for that looks rather ctyptic
again, but AFAIKS it deals mostly with checking for badblocks amogst
blocks being reserved. Since we do not support badblocks, the code can
be
Would you then make a kind of announce that mke2fs is considered frozen?
I think interested people will notice that we have mke2fs without
special announcement.
I see. I just meant to notify of settling the code down, so they can
eventually start testing.
Best regards,
--
Vladimir
The diff says inode count is wrong:
# ./mkfs_ext2_test.sh
Testing 583249
Testing 172
Testing 240769
Testing 465535
Testing 291723
Testing 29159
Testing 483732
Testing 541209
--- image_bb.out Tue Oct 20 23:32:39 2009
+++ image_std.out Tue Oct 20 23:32:39 2009
@@ -2,11
man mke2fs says that -I inode_size must be = 128
and must be power of 2. So, it can be 128, i.e.,
smaller than sizeof(struct ext2_inode_large), or 256, 512, etc,
i.e. sizeof(struct ext2_inode_large). So you just can check
inodesize != 128 instead.
You mean we can safely put 256-128 instead
Well, from the source it is not obvious at all. Is it this cryptic
rgdtsz stuff? Can it have a better name?
reserved group descriptor table size
Why should I hack vanilla mke2fs?
For both to play the same game. BB mke2fs right now doesn't reserve
group descriptors, thus wrong assumptions
* some clarifications in comments
* consistent /lost+found
Against current git. Please, consider applying.
--
Vladimir
test.patch
Description: Binary data
___
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
Against current git. Please, consider applying.
Sorry... I pushed some fixes to git...
can you rediff?
Sure. In a couple of minutes.
In current git, mkfs_ext2_test.sh shows that we are ok
in the whole 60k...8377k range, [8378..8410] is bad,
and I did not find yet the next bad value, but
1 - 100 of 414 matches
Mail list logo