Things to do this week instead of arguing about mixers
by Eric Sandeen
Now that we have ext4 as the new default filesystem, it'd be nice if we
can get more applications to take advantage of some of the features.
One big feature that has already been brought up on the list[1] is file
preallocation, which allows an application to pre-allocate blocks it
knows that it will eventually write into, thereby making sure it won't
run out of space, and also generally getting a more efficient/contiguous
file layout.
Only a few applications are taking advantage of this so far, in part
because it's new.[2] The transmission bittorrent client is using it,
but only if you tweak a configfile in (IMHO) non-obvious ways.
So it'd be great to help more applications take advantage of this, and
evangelize the interface a bit, and maybe do it in a semi-organized
fashion. First a bit of background on what this interface is:
Filesystems which can flag ranges of blocks as allocated but not
initialized can preallocate those blocks to a file very quickly, without
needing to write 0s or do any actual file data IO to perform the
allocation. When read, they return 0 because they are flagged as
uninitialized.
ext4 (as well as btrfs, ocfs2, and xfs) has an ->fallocate inode
operation which is the hook to this fast preallocation interface. The
low-level interface is via a syscall, but this is unlikely what we'd
like the applications to use directly. There are 2 other paths to the
interface:
posix_fallocate(3):
int posix_fallocate(int fd, off_t offset, off_t len);
This has existed for some time, but recent glibc will call the efficient
syscall if the underlying filesystem supports it. It will fall back to
essentially writing 0s to the file if not (or if on older glibc), and
this may not be desired behavior; preallocating 10G could take a very
long time this way.
fallocate(2):
long fallocate(int fd, int mode, loff_t offset, loff_t len);
This is directly wired to the syscall, so only succeeds on filesystems
that support it. It also takes a FALLOC_FL_KEEP_SIZE mode argument,
which allows one to allocate blocks without updating the file size if
desired (blocks can then be allocated past EOF). This call is only
wired up in very recent glibc, but it is available in F11.
So, tasks I see to be done, to get this started:
* Come up with some template autoconf magic to make it easy for
apps to detect fallocate() at build time, and some example
code on how to use it
- Should it fall back to posix_fallocate if fallocate is absent?
* Decide on some consistent buildt-time, run-time, and
configuration behavior when enabling this
- should build time use posix_fallocate if only it is available?
- config enabled == use fallocate whenever the fs supports it?
- config enabled == fall back to posix_fallocate or not?
- I'd be happy enough with exclusively using fallocate()
* Come up with a list of apps which could benefit:
- all torrent clients?
- rsync? (some patches have floated before)
- rpm? (file installation and/or db files?)
- databases?
- file downloaders?
- virt image tools?
- ____ ?
* Work with Fedora package maintainers and/or upstream to get this
hooked up where appropriate
* (Make a wiki page or a tracker bug to follow all this?)
Whaddya say? Anyone want to help with this?
Thanks,
-Eric
[1]https://www.redhat.com/archives/fedora-devel-list/2009-April/msg00110.html
[2]https://www.redhat.com/archives/fedora-devel-list/2009-April/msg02078.html
14 years, 12 months
[REPOST!] Split out e2fsprogs sublibraries
by Richard W.M. Jones
[I posted this before, but no one replied, so sending again to
fedora-devel-list]
https://bugzilla.redhat.com/show_bug.cgi?id=225406#c7
I would like to propose that e2fsprogs generate four subpackages for
the independent libraries that it contains. These four libraries are
used by other packages that don't need the whole of e2fsprogs-devel
(eg. krb5_workstation uses libss, qpid uses libuuid, and many programs
use libcom_err).
Our specific use case is to help with ongoing work porting libraries
to MinGW (http://fedoraproject.org/wiki/MinGW) where we would prefer
to package mingw32-libuuid for mingw32-qpidc without needing to port
the whole of e2fsprogs.
I looked at Debian's package, and would like to propose a split along
the same lines:
http://packages.debian.org/source/lenny/e2fsprogs
Despite the apparent complexity, there are only really four
subpackages. For the Fedora package we would create:
libblkid libblkid-devel
libcom_err libcom_err-devel [note 1]
libss libss-devel
libuuid libuuid-devel
There are no conflicting package names in Fedora at the moment, except
for the similarly named libss7 (a library implementing Signalling
System 7 telephone switching protocol).
I have attached a patch against Rawhide which does the above split. I
set up the dependencies so there should be no loss of functionality
for users who install just e2fsprogs or e2fsprogs-devel.
What remains is to advertize the split on fedora-devel-list and
encourage package maintainers to replace:
BuildRequires: e2fsprogs-devel
with
BuildRequires: lib<uuid|ss|blkid|com_err>-devel
where appropriate.
Rich.
--
Richard Jones, Emerging Technologies, Red Hat http://et.redhat.com/~rjones
New in Fedora 11: Fedora Windows cross-compiler. Compile Windows
programs, test, and build Windows installers. Over 70 libraries supprt'd
http://fedoraproject.org/wiki/MinGW http://www.annexia.org/fedora_mingw
14 years, 12 months
rawhide report: 20090512 changes
by Fedora compose checker
Compose started at Tue May 12 06:15:04 UTC 2009
Removed package cpan2rpm
Removed package kio_p7zip
Updated Packages:
docbook-dtds-1.0-47.fc11
------------------------
* Mon May 11 2009 Ondrej Vasik <ovasik(a)redhat.com> - 1.0-47
- Added requires(post) for /bin/chmod (#498680)
fedora-package-config-apt-11-4
------------------------------
* Tue May 05 2009 Axel Thimm <Axel.Thimm(a)ATrpms.net> - 11-4
- Update to F11.
fedora-package-config-smart-11-17
---------------------------------
* Tue May 05 2009 Axel Thimm <Axel.Thimm(a)ATrpms.net> - 11-17
- Switch to stable release.
fedora-release-11-1
-------------------
* Mon May 11 2009 Jesse Keating <jkeating(a)redhat.com> - 11-1
- First attempt at release candidate.
- Drop the compose configs, that's in spin-kickstarts
- Enable fedora and updates repo, disable rawhide
gallery2-2.3-9.fc11
-------------------
* Fri May 01 2009 Jon Ciesla <limb(a)jcomserv.net> - 2.3-9
- Add rewrite dep for httpauth, BZ 498061.
- Adopt rdieter's symlink handling suggestions from rel-eng #1674.
* Tue Apr 28 2009 Jon Ciesla <limb(a)jcomserv.net> - 2.3-8
- pretrans script logic fix. BZ498019.
giflib-4.1.6-2.fc11
-------------------
* Sat May 09 2009 Robert Scheck <robert(a)fedoraproject.org> 4.1.6-2
- Solved multilib problems with documentation (#465208, #474538)
- Removed static library from giflib-devel package (#225796 #c1)
glibc-2.10.1-1
--------------
* Sun May 10 2009 Jakub Jelinek <jakub(a)redhat.com> 2.10.1-1
- fix up getsgent_r and getsgnam_r exports on i?86 and ppc
* Sat May 09 2009 Jakub Jelinek <jakub(a)redhat.com> 2.10-2
- update from trunk
- glibc 2.10 release
- fix memchr on x86_64 (#499689)
leonidas-backgrounds-11.0.0-1.fc11
----------------------------------
* Sat May 09 2009 Martin Sourada <mso(a)fedoraproject.org> - 11.0.0-1
- Include the lion design optionally on single screens as well via -lion
subpackage
- Split the dual screen images with lion design into -lion-dual subpackage
to allow having only one of the single screens installed
- Don't forget to own some all used directories not provided by other pkgs
libvirt-0.6.2-8.fc11
--------------------
* Sun May 10 2009 Cole Robinson <crobinso(a)redhat.com> - 0.6.2-8.fc11
- Don't try to label a disk with no path (e.g. empty cdrom) (bug #499569)
* Thu May 07 2009 Mark McLoughlin <markmc(a)redhat.com> - 0.6.2-7.fc11
- Enable migration for qemu 0.10 (bug #499704)
mesa-7.5-0.14.fc11
------------------
* Tue May 05 2009 Dave Airlie <airlied(a)redhat.com> 7.5-0.14
- radeon-rewrite.patch: fixes from upstream for rs690 + r200
* Tue Apr 28 2009 Dave Airlie <airlied(a)redhat.com> 7.5-0.12
- rebase to upstream snapshot + radeon-rewrite
* Tue Apr 28 2009 Dave Airlie <airlied(a)redhat.com> 7.5-0.13
- radeon fix clip emits
* Thu Apr 16 2009 Dave Airlie <airlied(a)redhat.com> 7.5-0.11
- radeon-rewrite-fixes.patch: fix context crash in compiz + r200 fixes
* Tue Apr 14 2009 Adam Jackson <ajax(a)redhat.com> 7.5-0.10
- mesa-7.5-get-driver-name.patch: Fix glXGetScreenDriver for DRI2 (#495342)
mock-0.9.16-1.fc11
------------------
* Mon May 11 2009 Jesse Keating <jkeating(a)redhat.com> - 0.9.15-1
- Add configs for F11 (jkeating)
* Mon May 11 2009 Jesse Keating <jkeating(a)redhat.com> - 0.9.16-1
- Make F11 and rawhide build i586 on i386 targets.
selinux-policy-3.6.12-34.fc11
-----------------------------
* Mon May 11 2009 Dan Walsh <dwalsh(a)redhat.com> 3.6.12-34
- Allow rpcd_t to send signals to kernel threads
* Thu May 07 2009 Dan Walsh <dwalsh(a)redhat.com> 3.6.12-31
- Add policy for /var/lib/fprint
* Thu May 07 2009 Dan Walsh <dwalsh(a)redhat.com> 3.6.12-33
- Fix upgrade for F10 to F11
* Tue May 05 2009 Dan Walsh <dwalsh(a)redhat.com> 3.6.12-29
- Allow svirt to manage pci and other sysfs device data
* Tue May 05 2009 Dan Walsh <dwalsh(a)redhat.com> 3.6.12-30
-Remove duplicate line
valgrind-3.4.1-3
----------------
* Mon May 11 2009 Jakub Jelinek <jakub(a)redhat.com> 3.4.1-3
- rebuilt against glibc 2.10.1
xorg-x11-xinit-1.0.9-7.fc11
---------------------------
* Fri May 08 2009 Adam Jackson <ajax(a)redhat.com> 1.0.9-7
- xinit-1.0.9-unset.patch: Also unset XDG_SESSION_COOKIE in startx. (#489999)
Summary:
Added Packages: 0
Removed Packages: 2
Modified Packages: 14
Broken deps for ppc64
----------------------------------------------------------
cabal2spec-0.12-1.fc11.noarch requires ghc > 0:6.10.1-7
14 years, 12 months
Updated translations for gnome-packagekit in F-11 GA?
by Ville-Pekka Vainio
Hi,
As gnome-packagekit is the default graphical update tool on the GNOME
desktop, would it be possible to get a package with updated translations
to Fedora 11 GA version? There were quite a lot of changes during the
F-11 cycle, including the new update viewer, and because gpk isn't a
Fedora sub-project, it didn't go through the string and translation
freezes we have in the Fedora L10n project. It would be nice to have as
complete translations as possible in the released version of F-11.
This might also be the case for kpackagekit but I haven't taken part in
its translation so I don't know about the L10n situation there.
--
Ville-Pekka Vainio
14 years, 12 months
Locking assertion failure. Backtrace:
by Ankur Sinha
hi,
I'm packaging an app called "panini". The packaging's gone smoothly.. On
running the app, I get
[Ankur@Ankur twitter_tui]$ panini
Locking assertion failure. Backtrace:
#0 /usr/lib/libxcb-xlib.so.0 [0xf09767]
#1 /usr/lib/libxcb-xlib.so.0(xcb_xlib_lock+0x2e) [0xf0990e]
#2 /usr/lib/libX11.so.6 [0x5a90e9]
#3 /usr/lib/libX11.so.6(XCreateSimpleWindow+0x26) [0x57edb6]
#4 /usr/lib/libQtGui.so.4 [0x72c3c85]
#5 /usr/lib/libQtGui.so.4(_ZN14QWidgetPrivate10create_sysEmbb+0x1828)
[0x72c25d8]
#6 /usr/lib/libQtGui.so.4(_ZN7QWidget6createEmbb+0x16c) [0x7284c1c]
#7 /usr/lib/libQtGui.so.4(_ZN14QWidgetPrivate11createWinIdEm+0x1e3)
[0x727ff43]
#8 /usr/lib/libQtGui.so.4(_ZN14QWidgetPrivate21setWindowTitle_helperERK7QString+0x9b) [0x72845bb]
#9 /usr/lib/libQtGui.so.4(_ZN7QWidget14setWindowTitleERK7QString+0xa2)
[0x7284a12]
#10 /usr/lib/libQtGui.so.4(_ZN11QMessageBox14setWindowTitleERK7QString
+0x24) [0x7766554]
#11 /usr/lib/libQtGui.so.4 [0x7768237]
#12 /usr/lib/libQtGui.so.4(_ZN11QMessageBoxC1ENS_4IconERK7QStringS3_6QFlagsINS_14StandardButtonEEP7QWidgetS4_IN2Qt10WindowTypeEE+0x1f0) [0x77685b0]
#13 /usr/lib/libQtGui.so.4 [0x776885b]
#14 /usr/lib/libQtGui.so.4(_ZN11QMessageBox7warningEP7QWidgetRK7QStringS4_6QFlagsINS_14StandardButtonEES6_+0x36) [0x7768a76]
#15 panini [0x8057ca3]
#16 /usr/lib/libQtCore.so.4(_Z17qt_message_output9QtMsgTypePKc+0x35)
[0x6f0d095]
#17 /usr/lib/libQtCore.so.4(_Z8qWarningPKcz+0x71) [0x6f0d431]
#18 /usr/lib/libQtGui.so.4 [0x7290268]
#19 /usr/lib/libX11.so.6(_XError+0x109) [0x5a1aa9]
^\Quit
[Ankur@Ankur twitter_tui]$
I've googled and come across some posts related to the same, but they're
related to Java etc. and were solved by exporting some varialbles, which
did not work here. Can someone please tell me what the problem is?
regards,
Ankur
14 years, 12 months
yum-presto installed by default for Fedora 11?
by Rahul Sundaram
Hi,
Is the yum plugin going to be installed by default in Fedora 11? The
feature page says, no but I wanted to confirm this since there been a
number of changes back and forth related to delta rpms
Rahul
14 years, 12 months
Another update, another reboot, no sound!
by Dimi Paun
This is getting ridiculous!
I had running sound, and today I've updated my system.
Right away, my sound was b0rken:
- no sound from Skype, in any configuration imaginable
(even when ran through pasuspender)
- no sound from Flash in Firefox
- no sound from VirtualBox, even though I was seeing
the stream listed under "Playback" tab in PA's VC
- however, I was getting sound from Rhythmbox!
So I figure, I'll reboot, what else? Sure enough, after
reboot I get no sound at all! And this after fscking around
for about 30min(!) to get any sort of sound out of my system.
It turns out that I had to run "alsamixer" manually _again_
because the master volume was, logically, set to zero. And
this after filling a bug about this:
https://bugzilla.redhat.com/show_bug.cgi?id=493790
The bug was marked as fixed.
Sure enough, even after fixing the master volume, I still:
- have no sound from Skype in any imaginable combination,
even if it was working fine before (and I have not
changed the version of Skype that I use)
- no sound out of VirtualBox, even if it was working before
I've tried filing bugs, it doesn't help -- the rate of breakage
is way faster than I can file bugs (not to mention that filing
bugs is too sluggish for words).
Something is Just Plain Wrong (TM). WTH is going on? How can the
situation be improved? I've just lost 1h right now trying to deal
with this problem, in the middle of my work day -- this is not
sustainable.
--
Dimi Paun <dimi(a)lattica.com>
Lattica, Inc.
14 years, 12 months
rtorrent GCC 4.1 incompatibility?
by Michael Schwendt
[Bcc to rtorrent-owner]
The rtorrent.spec file does
| # work around a bug thats triggered by gcc 4.1
|
| export RPM_OPT_FLAGS=`echo $RPM_OPT_FLAGS | %{__sed} s/-O2/-Os/`
and the reader wonders what bug that may be?
No bugzilla ticket is linked. The %changelog says:
| * Sun Nov 26 2006 Chris Chabot <chabotc(a)xs4all.nl> - 0.6.4-1
| - New upstream version
| - Compile with -Os to work around a gcc 4.1 incompatibility
No bz number either. And it's an entry from 2006. We're in 2009,
however, and are using a much newer GCC release.
So, what GCC 4.1 bug is it? Is the -Os still needed?
14 years, 12 months
Note on man pages
by Rahul Sundaram
Hi,
Debian has a policy that all the commands should have man pages as part
of their review process. Unfortunately they don't seem to be sending it
upstream. If you are a package maintainer for some application, that is
missing a man page, it might be a good to look into the Debian sources
for man page patches, add them to your package and send it upstream as
well. Note that upstream might require you to clean up the man page in
the process.
Rahul
14 years, 12 months