Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
GCC 4.5 testing
View unanswered posts
View posts from last 24 hours

Goto page Previous  1, 2, 3 ... 9, 10, 11 ... 13, 14, 15  Next  
Reply to topic    Gentoo Forums Forum Index Unsupported Software
View previous topic :: View next topic  
Author Message
wrc1944
Advocate
Advocate


Joined: 15 Aug 2002
Posts: 3231
Location: Gainesville, Florida

PostPosted: Fri Aug 06, 2010 2:01 pm    Post subject: Reply with quote

kernelOfTruth,
Thanks for the tips and bug links. :)
Think I'll wait on my main box- can't afford to mess that one up.

BTW, when trying to enable graphite on an ~x86 box, I ran into a ppl and cloog problem- couldn't compile ppl, so I couldn't emerge gcc-4.5.1 with graphite. Worked OK without graphite enabled. However, on my pure ~amd64 box (identical Gentoo installation, just pure amd64, no-multilib), graphite did work. I didn't notice at the time, but emerging ppl must have worked OK, else gcc-4.5.1 wouldn't have installed.
_________________
Main box- AsRock x370 Gaming K4
Ryzen 1700, 3.0GHz, 16GB GSkill Flare DDR4 3200mhz
Samsung SATA 1000GB, Radeon HD R7 350 2GB DDR5
Gentoo ~amd64 plasma, glibc-2.29-r2, gcc-9.1.0 kernel-5.0.14-gentoo USE=experimental
Back to top
View user's profile Send private message
kernelOfTruth
Watchman
Watchman


Joined: 20 Dec 2005
Posts: 6108
Location: Vienna, Austria; Germany; hello world :)

PostPosted: Fri Aug 06, 2010 2:29 pm    Post subject: Reply with quote

wrc1944 wrote:
kernelOfTruth,
Thanks for the tips and bug links. :)
Think I'll wait on my main box- can't afford to mess that one up.

BTW, when trying to enable graphite on an ~x86 box, I ran into a ppl and cloog problem- couldn't compile ppl, so I couldn't emerge gcc-4.5.1 with graphite. Worked OK without graphite enabled. However, on my pure ~amd64 box (identical Gentoo installation, just pure amd64, no-multilib), graphite did work. I didn't notice at the time, but emerging ppl must have worked OK, else gcc-4.5.1 wouldn't have installed.


you're welcome :)

ah well, I should have been more clear:

you can upgrade & emerge stuff - but afterwards you (almost, since not all but most apps are affected) can't emerge stuff with the gcc-version minor than your current anymore [== it will fail]

e.g. switching from gcc-4.4.3 to gcc-4.5.1 via gcc-config and then you need to switch back to gcc-4.4.3 for certain apps --> no go :(
_________________
https://github.com/kernelOfTruth/ZFS-for-SystemRescueCD/tree/ZFS-for-SysRescCD-4.9.0
https://github.com/kernelOfTruth/pulseaudio-equalizer-ladspa

Hardcore Gentoo Linux user since 2004 :D
Back to top
View user's profile Send private message
wrc1944
Advocate
Advocate


Joined: 15 Aug 2002
Posts: 3231
Location: Gainesville, Florida

PostPosted: Fri Aug 06, 2010 3:23 pm    Post subject: Reply with quote

I don't suppose we could just add this to xibo's ebuild and re-digest, could we? :roll: Probably too much to hope for, I guess.
PATCH_VER="1.4"
UCLIBC_VER="1.0"

Or maybe just remame the portage ebuild to 4.5.1, and comment out the sed line near the bottom, and re-digest?

I'm sure this reveals my relative ignorance. I did look in the files directory in sys-devel/gcc, but didn't find any PATCH_VER="1.4" item, so I'm not sure of where the 4.5.0 ebuild gets the 1.4 patch, unless it downloads it. Probably a problem applying it to 4.5.1?
_________________
Main box- AsRock x370 Gaming K4
Ryzen 1700, 3.0GHz, 16GB GSkill Flare DDR4 3200mhz
Samsung SATA 1000GB, Radeon HD R7 350 2GB DDR5
Gentoo ~amd64 plasma, glibc-2.29-r2, gcc-9.1.0 kernel-5.0.14-gentoo USE=experimental
Back to top
View user's profile Send private message
kernelOfTruth
Watchman
Watchman


Joined: 20 Dec 2005
Posts: 6108
Location: Vienna, Austria; Germany; hello world :)

PostPosted: Fri Aug 06, 2010 4:51 pm    Post subject: Reply with quote

wrc1944 wrote:
I don't suppose we could just add this to xibo's ebuild and re-digest, could we? :roll: Probably too much to hope for, I guess.
PATCH_VER="1.4"
UCLIBC_VER="1.0"

Or maybe just remame the portage ebuild to 4.5.1, and comment out the sed line near the bottom, and re-digest?

I'm sure this reveals my relative ignorance. I did look in the files directory in sys-devel/gcc, but didn't find any PATCH_VER="1.4" item, so I'm not sure of where the 4.5.0 ebuild gets the 1.4 patch, unless it downloads it. Probably a problem applying it to 4.5.1?


:roll:

just go ahead and try ;)

nope - this is a tarball, which gets downloaded from the mirrors, :)

most of the patches can be re-used, some need to be removed since 4.5.1 already includes the fixes of them

I already created a 1.0.1 patchset - based on the 1.4 patchset of 4.5.0 - for 4.5.1 unfortunately the most crucial patch (namely the above mentioned fix for the "cloog disaster" - everything being linked against it didn't apply cleanly

so if you are/were not dependent on that you easily could bump it with gentoo's additional patches

the current state (the failing patch) doesn't make it that easy

I'm sure dirtyepic or other devs are already working on it or will soon :idea:
_________________
https://github.com/kernelOfTruth/ZFS-for-SystemRescueCD/tree/ZFS-for-SysRescCD-4.9.0
https://github.com/kernelOfTruth/pulseaudio-equalizer-ladspa

Hardcore Gentoo Linux user since 2004 :D
Back to top
View user's profile Send private message
ComaWhite
Tux's lil' helper
Tux's lil' helper


Joined: 07 Oct 2008
Posts: 125

PostPosted: Sun Aug 08, 2010 12:11 am    Post subject: Reply with quote

What's the state of 4.5.1 with Gentoo with just simple flags like -O2 -pipe -march=native? Will I have issues like you guys or what?
Back to top
View user's profile Send private message
kernelOfTruth
Watchman
Watchman


Joined: 20 Dec 2005
Posts: 6108
Location: Vienna, Austria; Germany; hello world :)

PostPosted: Sun Aug 08, 2010 12:23 am    Post subject: Reply with quote

ComaWhite wrote:
What's the state of 4.5.1 with Gentoo with just simple flags like -O2 -pipe -march=native? Will I have issues like you guys or what?


if gcc 4.5.0 worked fine so far for you, 4.5.1 should be no problem :wink:
_________________
https://github.com/kernelOfTruth/ZFS-for-SystemRescueCD/tree/ZFS-for-SysRescCD-4.9.0
https://github.com/kernelOfTruth/pulseaudio-equalizer-ladspa

Hardcore Gentoo Linux user since 2004 :D
Back to top
View user's profile Send private message
rhill
Retired Dev
Retired Dev


Joined: 22 Oct 2004
Posts: 1629
Location: sk.ca

PostPosted: Sun Aug 08, 2010 2:07 am    Post subject: Reply with quote

4.5.1 is in the tree.
_________________
by design, by neglect
for a fact or just for effect
Back to top
View user's profile Send private message
devsk
Advocate
Advocate


Joined: 24 Oct 2003
Posts: 2860
Location: Bay Area, CA

PostPosted: Sun Aug 08, 2010 2:31 am    Post subject: Reply with quote

Is there a package in portage which doesn't compile and/or run with gcc-4.5.1 with minimal c[xx]flags like "-O2"? I think rephrased, the question is: is 4.5.1 safe enough for normal usage (forget about exotic graphite/lto)?
Back to top
View user's profile Send private message
rhill
Retired Dev
Retired Dev


Joined: 22 Oct 2004
Posts: 1629
Location: sk.ca

PostPosted: Sun Aug 08, 2010 3:24 am    Post subject: Reply with quote

You can always check the GCC 4.5 porting tracker.

https://bugs.gentoo.org/showdependencytree.cgi?id=296658&hide_resolved=1

When the few remaining big ticket items (ie. MySQL, emacs) are fixed then we'll unmask 4.5.1.
_________________
by design, by neglect
for a fact or just for effect
Back to top
View user's profile Send private message
ComaWhite
Tux's lil' helper
Tux's lil' helper


Joined: 07 Oct 2008
Posts: 125

PostPosted: Sun Aug 08, 2010 4:05 am    Post subject: Reply with quote

Well then once mysql-5.1.49 gets fixed for GCC-4.5.x then I'll make the switch the rest of the stuff like emacs I don't care for (maybe audacity but idk).
Back to top
View user's profile Send private message
Etal
Veteran
Veteran


Joined: 15 Jul 2005
Posts: 1682

PostPosted: Sun Aug 08, 2010 4:33 am    Post subject: Reply with quote

Wasn't there some nasty Tcl bug with the GCC 4.5.0? I can't seem to find it on the tracker

Edit: This one (it should probably block the tracker?)
Back to top
View user's profile Send private message
devsk
Advocate
Advocate


Joined: 24 Oct 2003
Posts: 2860
Location: Bay Area, CA

PostPosted: Sun Aug 08, 2010 4:52 am    Post subject: Reply with quote

So, basically I should just inject (package.provided) the failing packages while updating the world with 4.5.1, and I should be fine. I hope if I have older GCC installed, the already emerged packages will continue to work.
Back to top
View user's profile Send private message
devsk
Advocate
Advocate


Joined: 24 Oct 2003
Posts: 2860
Location: Bay Area, CA

PostPosted: Sun Aug 08, 2010 4:54 am    Post subject: Reply with quote

Code:
# emerge -pv @preserved-rebuild

These are the packages that would be merged, in order:

Calculating dependencies ... done!

emerge: there are no ebuilds to satisfy "sys-devel/gcc:x86_64-pc-linux-gnu-4.4.4".
Does anybody know what this means?
Back to top
View user's profile Send private message
rhill
Retired Dev
Retired Dev


Joined: 22 Oct 2004
Posts: 1629
Location: sk.ca

PostPosted: Sun Aug 08, 2010 5:31 am    Post subject: Reply with quote

AM088 wrote:
Wasn't there some nasty Tcl bug with the GCC 4.5.0? I can't seem to find it on the tracker

Edit: This one (it should probably block the tracker?)


Thanks, I hadn't seen that.

devsk wrote:
So, basically I should just inject (package.provided) the failing packages while updating the world with 4.5.1, and I should be fine. I hope if I have older GCC installed, the already emerged packages will continue to work.


For mysql at least I just masked >=dev-db/mysql-5.1 and >=virtual/mysql-5.1. Older versions work.

Quote:
Does anybody know what this means?


It thinks you have gcc installed in a x86_64-pc-linux-gnu-4.4.4 SLOT. Do you have toolchains built by crossdev installed or USE=multislot enabled? This sounds like bug #247581
_________________
by design, by neglect
for a fact or just for effect
Back to top
View user's profile Send private message
devsk
Advocate
Advocate


Joined: 24 Oct 2003
Posts: 2860
Location: Bay Area, CA

PostPosted: Sun Aug 08, 2010 5:45 am    Post subject: Reply with quote

dirtyepic wrote:
AM088 wrote:
Wasn't there some nasty Tcl bug with the GCC 4.5.0? I can't seem to find it on the tracker

Edit: This one (it should probably block the tracker?)


Thanks, I hadn't seen that.

devsk wrote:
So, basically I should just inject (package.provided) the failing packages while updating the world with 4.5.1, and I should be fine. I hope if I have older GCC installed, the already emerged packages will continue to work.


For mysql at least I just masked >=dev-db/mysql-5.1 and >=virtual/mysql-5.1. Older versions work.

Quote:
Does anybody know what this means?


It thinks you have gcc installed in a x86_64-pc-linux-gnu-4.4.4 SLOT. Do you have toolchains built by crossdev installed or USE=multislot enabled? This sounds like bug #247581
I have USE=multislot. I thought that was required for keeping multiple versions of GCC and binutils around. Looks like its not so. Can you please confirm, dirty?
Back to top
View user's profile Send private message
devsk
Advocate
Advocate


Joined: 24 Oct 2003
Posts: 2860
Location: Bay Area, CA

PostPosted: Sun Aug 08, 2010 5:47 am    Post subject: Reply with quote

Changing this would mean that I need to compile binutils and all existing versions of gcc with USE=-multislot. Ouch!
Back to top
View user's profile Send private message
rhill
Retired Dev
Retired Dev


Joined: 22 Oct 2004
Posts: 1629
Location: sk.ca

PostPosted: Sun Aug 08, 2010 6:17 am    Post subject: Reply with quote

devsk wrote:
I have USE=multislot. I thought that was required for keeping multiple versions of GCC and binutils around. Looks like its not so. Can you please confirm, dirty?


It's for keeping every intermediate version installed. Normally gcc is installed in a SLOT like "4.5". With multislot it's installed into something like "${CTARGET}-4.5.0" so both 4.5.0 and 4.5.1 can be installed simultaneously. However this rewriting of SLOT on the fly causes all kinds of havok because the SLOT that gets recorded in the world file doesn't match any ebuild in the tree as far as portage can tell.

In short, don't use it unless you have a very compelling reason for needing every last version installed.
_________________
by design, by neglect
for a fact or just for effect
Back to top
View user's profile Send private message
devsk
Advocate
Advocate


Joined: 24 Oct 2003
Posts: 2860
Location: Bay Area, CA

PostPosted: Sun Aug 08, 2010 7:53 am    Post subject: Reply with quote

python speed up by 5% with 4.5.1. I haven't built glibc with 4.5.1 yet, just python. Can others try and reproduce:

Code:
time python -c "import math ; l = [i + i for i in xrange(100000000)];"
Run it a couple of times and compare.

time emerge -puDvN world, on fully loaded cache (i.e. disk is no longer the bottleneck i.e. this was run twice) drops by 4%.
Back to top
View user's profile Send private message
xibo
Apprentice
Apprentice


Joined: 21 Aug 2007
Posts: 152
Location: moving between kubuntu and ubuntu kde edition

PostPosted: Sun Aug 08, 2010 12:35 pm    Post subject: Reply with quote

ffmpeg and gzip also run faster.
[rant]
Nepomuk is still slow though :P
[/rant]
Back to top
View user's profile Send private message
wrc1944
Advocate
Advocate


Joined: 15 Aug 2002
Posts: 3231
Location: Gainesville, Florida

PostPosted: Sun Aug 08, 2010 1:36 pm    Post subject: Reply with quote

Previously, I used xibo's 4.5.1 ebuild on my amd64 box (graphite enabled globally), and it's working OK. Do I need to rebuild using the new portage ebuild 4.5.1?

Xibo had the sed -i 's/use_fixproto=yes/:/' gcc/config.gcc #PR33200 line commented out, and PATCH_VER="1.4", whereas the portaged PR33200 is not commented, and it uses PATCH_VER="1.0".

I assume graphite is still not recommended, and the ppl/cloog problem is still there. Correct? But if that's true, how is my amd64 box (full @world rebuild) using it? dev-libs/ppl-0.10.2-r1:0 compiled installed OK, and cloog-ppl too. Is the problem only on ~x86?

Really looking forward to moving my main ~x86 box to 4.5.1. :D
_________________
Main box- AsRock x370 Gaming K4
Ryzen 1700, 3.0GHz, 16GB GSkill Flare DDR4 3200mhz
Samsung SATA 1000GB, Radeon HD R7 350 2GB DDR5
Gentoo ~amd64 plasma, glibc-2.29-r2, gcc-9.1.0 kernel-5.0.14-gentoo USE=experimental
Back to top
View user's profile Send private message
xibo
Apprentice
Apprentice


Joined: 21 Aug 2007
Posts: 152
Location: moving between kubuntu and ubuntu kde edition

PostPosted: Sun Aug 08, 2010 2:13 pm    Post subject: Reply with quote

the last post of the PR33200 entry of gcc bug database tells you wether you need that statement or not. You can keep the sed statement, it won't waste anything ( other then time ) though.
PATCH_VER is a gentoo-developed thing and since it's "value" was modified i guess a maintainer has taken a look at it for the 4.5.1 ebuild.

I don't know about graphite, but gcc ships it's own tests, and you can have emerge validate the gcc being emerged works (as intended by the gcc devs) by enabling the test use-flag.
Back to top
View user's profile Send private message
wrc1944
Advocate
Advocate


Joined: 15 Aug 2002
Posts: 3231
Location: Gainesville, Florida

PostPosted: Sun Aug 08, 2010 2:36 pm    Post subject: Reply with quote

Thanks xibo,
I figured it was not needed now, but wondered why dirtyepic left it in the 4.5.1 ebuild. Just an oversight, or is it needed in some cases, or with some hardware- just don't know. :?

I'm trying to understand if that PATCH_VER="1.4" version in your ebuild had something to do with ppl and cloog-ppl installing, which allowed me to get gcc-4.5.1 installed with graphite on amd64.

Then the portage ebuild has PATCH_VER="1.0", which I assume doesn't have any ppl/cloog fixes, thus won't let graphite be enabled? Maybe my current understanding is all wrong. Guess I'll try the portage ebuild on my main ~x86 box (with and without graphite), and see what happens. Maybe the 1.0 patch version has been revised? At this point it's a mystery to me. :roll:
_________________
Main box- AsRock x370 Gaming K4
Ryzen 1700, 3.0GHz, 16GB GSkill Flare DDR4 3200mhz
Samsung SATA 1000GB, Radeon HD R7 350 2GB DDR5
Gentoo ~amd64 plasma, glibc-2.29-r2, gcc-9.1.0 kernel-5.0.14-gentoo USE=experimental
Back to top
View user's profile Send private message
rhill
Retired Dev
Retired Dev


Joined: 22 Oct 2004
Posts: 1629
Location: sk.ca

PostPosted: Sun Aug 08, 2010 5:54 pm    Post subject: Reply with quote

No, it's a new version so the patchset gets reset to 1.0. It has all the patches 4.5.0 had.
_________________
by design, by neglect
for a fact or just for effect
Back to top
View user's profile Send private message
devsk
Advocate
Advocate


Joined: 24 Oct 2003
Posts: 2860
Location: Bay Area, CA

PostPosted: Sun Aug 08, 2010 6:28 pm    Post subject: Reply with quote

More datapoints: in-memory gzip is 5% faster and in-memory gunzip is 11% faster. No effect on bzip2. 0.5% slow down in LZMA...:)
Back to top
View user's profile Send private message
ssteinberg
Apprentice
Apprentice


Joined: 09 Jul 2010
Posts: 206
Location: Israel

PostPosted: Sun Aug 08, 2010 6:34 pm    Post subject: Reply with quote

-lto fails on git dbus as well. gcc 4.5.1
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Unsupported Software All times are GMT
Goto page Previous  1, 2, 3 ... 9, 10, 11 ... 13, 14, 15  Next
Page 10 of 15

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum