Re: ports/128754: [port infrastructure] implement master sites randomization
On Tue, Nov 11, 2008 at 03:23:50AM +, RW wrote: On Mon, 10 Nov 2008 18:56:16 +0300 (MSK) Eygene Ryabinkin [EMAIL PROTECTED] wrote: Today I was hit by the very bad connectivity with twaren.net and nchc.dl.sourceforge.net (the first site in the SF mirrors list in the FreeBSD ports .mk files) is hosted by Taiwan REN. So, I decided to implement simple randomization that will enable to evenly distribute the downloads between SF mirrors. ... +# Need to drop a couple of initial rand() values: they tend +# to be around 0.8 - 0.9, so for fairly small array lenght +# they will produce identical values at the beginning. + srand(); rand(); rand(); rand(); rand(); I think it would be sensible to seed srand from a hash of something reproducible to make better use of caches - maybe DISTNAME+DISTVERSION. Maybe I don't understand something, but is RANDOMIZE_MASTER_SITES (see bsd.port.mk for details) not enough? It affects though all sites, not only SF. Just my 0.02$, Alexey. ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: ports/128754: [port infrastructure] implement master sites randomization
Tue, Nov 11, 2008 at 04:35:54PM +0100, Alexey Shuvaev wrote:
On Tue, Nov 11, 2008 at 03:23:50AM +, RW wrote:
On Mon, 10 Nov 2008 18:56:16 +0300 (MSK)
I think it would be sensible to seed srand from a hash of something
reproducible to make better use of caches - maybe DISTNAME+DISTVERSION.
Maybe I don't understand something, but is RANDOMIZE_MASTER_SITES
(see bsd.port.mk for details) not enough? It affects though all
sites, not only SF.
It is enough, but it requires /usr/games/random to be present. Not all
servers have 'games' distribution installed, but I want them to
randomize master sites.
For the feeding the hashes: RW, do you mean HTTP caches? In principle,
this is a neat idea: it will achieve load-balancing between the sites.
But as it will use the same master sites order for the given port, this
will be failing when the first download site is almost down: the
download will take very long. But probably stable order of the sites
can be made settable via the variable, e.g.
RANDOMIZE_MASTER_SITE_REPRODUCIBLY. Will it be fine? Please, note that
this can be achievable only for the awk script: random(6) can not be
currently directed to do this.
--
Eygene
____ _.--. #
\`.|\.....-'` `-._.-'_.-'` # Remember that it is hard
/ ' ` , __.--' # to read the on-line manual
)/' _/ \ `-_, /# while single-stepping the kernel.
`-' `\_ ,_.-;_.-\_ ', fsc/as #
_.-'_./ {_.' ; / #-- FreeBSD Developers handbook
{_.-``-' {_/#
pgp7toHGdAGHq.pgp
Description: PGP signature
Re: Another VIA UniChrome/Chrome9 Driver
Robert Noland wrote: How does this compare to our Xorg driver? I've contemplated finishing the via drm work, but I don't have any of these devices and I'm not sure how many people do. At least that care about drm anyway... robert. I haven't look in it carefully. But skimming tells me that the official driver has more branches for various chips. See the source package: http://linux.via.com.tw/ Via_drv.so is actually a Linux ELF ABI. Only because it wasn't linked against any Linux's shared object can it run under FreeBSD simply. Otherwise, a mechanism like linuxpluginwrapper (Wrapper for Linux Mozilla/Firefox plugin to work with FreeBSD's native ABI of Mozilla/Firefox) must be used. The driver package contains some Linux kernel modules and shared objects, including libGL.so.1.2.via_chrome9 and via_chrome9_dri.so. Thus, I guess it is not easy to make the driver function fully under FreeBSD, especially 2D/3D acceleration. But so far, as I know, it's the unique solution that makes my laptop work. And I guess it supports more new chips. I have quite a few products with VIA chipset: * A laptop with UniChrome Pro IGP (0x3343) * Motherboard VIA PC2500E with north bridge CN700 * Motherboard VIA PC3500E with north bridge CN896 I can make tests with you. From Beijing, China ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Another VIA UniChrome/Chrome9 Driver
On Wed, 2008-11-12 at 01:02 +0800, Intron is my alias on the Internet wrote: But so far, as I know, it's the unique solution that makes my laptop work. What do you mean by work. Does the normal xorg driver not work for your chip? Let me go look at the ddx driver... And I guess it supports more new chips. I have quite a few products with VIA chipset: * A laptop with UniChrome Pro IGP (0x3343) This pci_id is listed in drm, so if I finish porting the code, it should work, for some definition of work... ;) robert. * Motherboard VIA PC2500E with north bridge CN700 * Motherboard VIA PC3500E with north bridge CN896 signature.asc Description: This is a digitally signed message part
Another VIA UniChrome/Chrome9 Driver
Port:http://ftp.intron.ac/tmp/xf86-video-via-linux.tar.gz If the distfile cannot be downloaded from VIA's website, please download it from my server: http://ftp.intron.ac/tmp/chrome9.83-242-sl10.1.tar.gz This is official driver from VIA, originally for SuSE 10.1. It works on my laptop (BENQ Joybook R41E) and VIA pc-1 motherboard. The GPU of my laptop is: [EMAIL PROTECTED]:1:0:0:class=0x03 card=0x86241071 chip=0x33431106 rev=0x01 hdr=0x00 vendor = 'VIA Technologies Inc' device = 'VIA/S3G UniChromePro IGP' class = display subclass = VGA OpenChrome doesn't work with it (chip's PID = 0x3343) at all. It supports features as well as OpenChrome: * XVideo * Rotation * Dual Display - Mono display on only either LCD or VGA only - DuoView (The same scene on both LCD and VGA simultaneously) I failed to get SAMM (different scenes on LCD and VGA). I believe it is useful for people who is using laptop with VIA chipset. And I hope it can be put into port tree before it is out of date. Any FreeBSD committee can help me? From Beijing, China ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Another VIA UniChrome/Chrome9 Driver
On Wed, 2008-11-12 at 03:42 +0800, Intron is my alias on the Internet wrote: Robert Noland wrote: On Wed, 2008-11-12 at 01:02 +0800, Intron is my alias on the Internet wrote: But so far, as I know, it's the unique solution that makes my laptop work. What do you mean by work. Does the normal xorg driver not work for your chip? Let me go look at the ddx driver... And I guess it supports more new chips. I have quite a few products with VIA chipset: * A laptop with UniChrome Pro IGP (0x3343) This pci_id is listed in drm, so if I finish porting the code, it should work, for some definition of work... ;) robert. * Motherboard VIA PC2500E with north bridge CN700 * Motherboard VIA PC3500E with north bridge CN896 You may see the mysterious phenomenon here: http://ftp.intron.ac/tmp/openchrome.avi (1.9MB) Is that with the xorg driver? An xorg.log might be helpful. robert. Play the video file with MPlayer. My laptop + OpenChrome 0.2.903 = A space alien's visit ? From Beijing, China ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-x11 To unsubscribe, send any mail to [EMAIL PROTECTED] signature.asc Description: This is a digitally signed message part
Re: CFT: astro/boinc-setiathome-enhanced update
Rene Ladan schreef: 123456789012345678901234567890123456789012345678901234567890123456789012 Ignore this ruler ;) Rene -- http://www.rene-ladan.nl/ GPG fingerprint = E738 5471 D185 7013 0EE0 4FC8 3C1D 6F83 12E1 84F6 (subkeys.pgp.net) ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Another VIA UniChrome/Chrome9 Driver
On Tue, 2008-11-11 at 22:53 +0800, Intron is my alias on the Internet wrote: Port:http://ftp.intron.ac/tmp/xf86-video-via-linux.tar.gz If the distfile cannot be downloaded from VIA's website, please download it from my server: http://ftp.intron.ac/tmp/chrome9.83-242-sl10.1.tar.gz This is official driver from VIA, originally for SuSE 10.1. It works on my laptop (BENQ Joybook R41E) and VIA pc-1 motherboard. The GPU of my laptop is: [EMAIL PROTECTED]:1:0:0:class=0x03 card=0x86241071 chip=0x33431106 rev=0x01 hdr=0x00 vendor = 'VIA Technologies Inc' device = 'VIA/S3G UniChromePro IGP' class = display subclass = VGA OpenChrome doesn't work with it (chip's PID = 0x3343) at all. It supports features as well as OpenChrome: * XVideo * Rotation * Dual Display - Mono display on only either LCD or VGA only - DuoView (The same scene on both LCD and VGA simultaneously) I failed to get SAMM (different scenes on LCD and VGA). I believe it is useful for people who is using laptop with VIA chipset. And I hope it can be put into port tree before it is out of date. Any FreeBSD committee can help me? How does this compare to our Xorg driver? I've contemplated finishing the via drm work, but I don't have any of these devices and I'm not sure how many people do. At least that care about drm anyway... robert. From Beijing, China ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to [EMAIL PROTECTED] signature.asc Description: This is a digitally signed message part
CFT: astro/boinc-setiathome-enhanced update
Hi, 123456789012345678901234567890123456789012345678901234567890123456789012 I have a patch[1] ready for the astro/boinc-setiathome-enhanced port to update it to 6.03 (which is the same as 5.27 but with the science and graphics code separated), and to add astropulse 4.28 (same as 4.35, but the version number never got updated in SVN) to it. To test it, download the patch and replace the contents of /usr/ports/astro/boinc-setiathome-enhanced with the contents of the patch. It builds on my 7.0-RELEASE amd64 box, but the resulting astropulse binary crashes at startup. This might have something to do with a polluted source directory for the setiathome part, as the source code for astropulse borrows from the setiathome source code. I have an astropulse test work unit available. The port fails to build on a 7.1-BETA2 i386 box at work. At some point during the configuration of astropulse, configure complains about a bad fd number (gurus around?). This happens both with the regenerated configure script and with the original one (remove './_autosetup' from line 78 of the Makefile). Thanks for any ideas, Rene [1] : ftp://rene-ladan.nl/pub/freebsd/boinc-setiathome-enhanced.tar.bz2 SIZE=2740 MD5=8bb6be7ad07e693d3e72cf49ff1e480f -- http://www.rene-ladan.nl/ GPG fingerprint = E738 5471 D185 7013 0EE0 4FC8 3C1D 6F83 12E1 84F6 (subkeys.pgp.net) ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Another VIA UniChrome/Chrome9 Driver
Robert Noland wrote: You may see the mysterious phenomenon here: http://ftp.intron.ac/tmp/openchrome.avi (1.9MB) Is that with the xorg driver? An xorg.log might be helpful. robert. Play the video file with MPlayer. My laptop + OpenChrome 0.2.903 = A space alien's visit ? The mysterious phenomenon is produced with the port xf86-video-openchrome-0.2.903. Here is the log file: http://ftp.intron.ac/tmp/openchrome-Xorg.0.log It ended when I pressed Ctrl-Alt-BackSpace. From Beijing, China ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: ports/128754: [port infrastructure] implement master sites randomization
On Tue, Nov 11, 2008 at 07:19:03PM +0300, Eygene Ryabinkin wrote: Tue, Nov 11, 2008 at 04:35:54PM +0100, Alexey Shuvaev wrote: On Tue, Nov 11, 2008 at 03:23:50AM +, RW wrote: On Mon, 10 Nov 2008 18:56:16 +0300 (MSK) I think it would be sensible to seed srand from a hash of something reproducible to make better use of caches - maybe DISTNAME+DISTVERSION. Maybe I don't understand something, but is RANDOMIZE_MASTER_SITES (see bsd.port.mk for details) not enough? It affects though all sites, not only SF. It is enough, but it requires /usr/games/random to be present. Not all servers have 'games' distribution installed, but I want them to randomize master sites. For the feeding the hashes: RW, do you mean HTTP caches? In principle, this is a neat idea: it will achieve load-balancing between the sites. But as it will use the same master sites order for the given port, this will be failing when the first download site is almost down: the download will take very long. But probably stable order of the sites can be made settable via the variable, e.g. RANDOMIZE_MASTER_SITE_REPRODUCIBLY. Will it be fine? Please, note that this can be achievable only for the awk script: random(6) can not be currently directed to do this. As a side-note to your discussion: If your concern is to pick sites which are fast for you, you may want to look at ports-mgmt/fastest_sites. It handles almost all MASTER_SITE_FOO definitions and is designed to sort based upon round-trip time for the TCP handshake to complete (best guess for least effort). The output is suitable for inclusion into make.conf. -- WXS ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: Another VIA UniChrome/Chrome9 Driver
On Wed, 2008-11-12 at 04:21 +0800, Intron is my alias on the Internet wrote: Robert Noland wrote: You may see the mysterious phenomenon here: http://ftp.intron.ac/tmp/openchrome.avi (1.9MB) Is that with the xorg driver? An xorg.log might be helpful. robert. Play the video file with MPlayer. My laptop + OpenChrome 0.2.903 = A space alien's visit ? The mysterious phenomenon is produced with the port xf86-video-openchrome-0.2.903. Here is the log file: Ok, the more I dig into this, the more confused I get... This driver is not from via... It is open source, but not developed with xorg... Give me a few minutes to see how things look and maybe we can just make a port from the source... robert. http://ftp.intron.ac/tmp/openchrome-Xorg.0.log It ended when I pressed Ctrl-Alt-BackSpace. From Beijing, China ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-x11 To unsubscribe, send any mail to [EMAIL PROTECTED] signature.asc Description: This is a digitally signed message part
Re: Another VIA UniChrome/Chrome9 Driver
On Tue, 2008-11-11 at 16:00 -0500, Robert Noland wrote: On Wed, 2008-11-12 at 04:21 +0800, Intron is my alias on the Internet wrote: Robert Noland wrote: You may see the mysterious phenomenon here: http://ftp.intron.ac/tmp/openchrome.avi (1.9MB) Is that with the xorg driver? An xorg.log might be helpful. robert. Play the video file with MPlayer. My laptop + OpenChrome 0.2.903 = A space alien's visit ? The mysterious phenomenon is produced with the port xf86-video-openchrome-0.2.903. Here is the log file: Ok, the more I dig into this, the more confused I get... This driver is not from via... It is open source, but not developed with xorg... Give me a few minutes to see how things look and maybe we can just make a port from the source... Ok, I need more coffee... I have all the versions sorted out now... So your are using xf86-video-openchrome from ports... It looks like there is an internal conflict about whether this chip is actually a P4M890 or a P4M900. I'm not sure which way to go, so we will try the 890 first... Please try the following patch to the openchrome port. http://people.freebsd.org/~rnoland/openchrome.patch robert. robert. http://ftp.intron.ac/tmp/openchrome-Xorg.0.log It ended when I pressed Ctrl-Alt-BackSpace. From Beijing, China ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-x11 To unsubscribe, send any mail to [EMAIL PROTECTED] -- Robert Noland [EMAIL PROTECTED] 2Hip Networks signature.asc Description: This is a digitally signed message part
Re: Another VIA UniChrome/Chrome9 Driver
On Tue, 2008-11-11 at 18:04 -0500, Robert Noland wrote: On Tue, 2008-11-11 at 16:00 -0500, Robert Noland wrote: On Wed, 2008-11-12 at 04:21 +0800, Intron is my alias on the Internet wrote: Robert Noland wrote: You may see the mysterious phenomenon here: http://ftp.intron.ac/tmp/openchrome.avi (1.9MB) Is that with the xorg driver? An xorg.log might be helpful. robert. Play the video file with MPlayer. My laptop + OpenChrome 0.2.903 = A space alien's visit ? The mysterious phenomenon is produced with the port xf86-video-openchrome-0.2.903. Here is the log file: Ok, the more I dig into this, the more confused I get... This driver is not from via... It is open source, but not developed with xorg... Give me a few minutes to see how things look and maybe we can just make a port from the source... Ok, I need more coffee... I have all the versions sorted out now... So your are using xf86-video-openchrome from ports... It looks like there is an internal conflict about whether this chip is actually a P4M890 or a P4M900. I'm not sure which way to go, so we will try the 890 first... Please try the following patch to the openchrome port. http://people.freebsd.org/~rnoland/openchrome.patch Ok, I spoke to a couple of the unichrome developers and my patch isn't quite right, but it is close enough for you to test. It seems that the openchrome driver needs to match not only the chip id, but the card id also. It apparently needs this to know what ports to enable (panel, crt). Both your chip id and board were there, but not in the right combination. I have a proper patch for your board now, which has also been committed to openchromes svn repo... robert. http://ftp.intron.ac/tmp/openchrome-Xorg.0.log It ended when I pressed Ctrl-Alt-BackSpace. From Beijing, China ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-x11 To unsubscribe, send any mail to [EMAIL PROTECTED] signature.asc Description: This is a digitally signed message part
Re: Another VIA UniChrome/Chrome9 Driver
Robert Noland wrote: Ok, I spoke to a couple of the unichrome developers and my patch isn't quite right, but it is close enough for you to test. It seems that the openchrome driver needs to match not only the chip id, but the card id also. It apparently needs this to know what ports to enable (panel, crt). Both your chip id and board were there, but not in the right combination. I have a proper patch for your board now, which has also been committed to openchromes svn repo... robert. Thank you for your patience. With your patch http://www.openchrome.org/trac/changeset/686/trunk/src, Xorg can work on my laptop one-off. That is, for the first time I start Xorg, it works well. For the second time, my laptop go to die. Maybe it is because I'm using 8.0-CURRENT in order to debug my USB drivers. And even for the first time, HWCursor doesn't work, which makes me have to enable SWCursor. Just before I read some codes in openchrome, xf86-video-via and the one on http://linux.via.com.tw/. I guess in the past years, VIA and OpenChrome probably borrowed codes from each other. Now, I propose that the driver from linux.via.com.tw needs to be pushed into the port tree of FreeBSD. After all, it is an alternative choice for users. I adore you and other developpers of OpenChrome very much. But after all you are doing hacking and/or reverse engineering. And the information about VIA chips that you have got is seriously inequal to what VIA know about itself. By the way, I have tested MPEG-2 decoding of VIA chips on all products in my hands (PM890, CN700 and CN896). The result disappointed me very much. I put much hope on CN896, the best one of them. I found Fedora Core 5 after long searching on the Internet here and there, and then configured Xorg with the driver from VIA (the last one on the page): http://www.viaarena.com/default.aspx?PageID=420OSID=30CatID=2770SubCatID=189 This driver is the only one that fully supports MPEG-2 decoder as far as I found. After compiling VeMP (http://sourceforge.net/projects/vemp/), I wished CN896 could play HDTV (1920x1080) MPEG-2 materials smoothly. But actually, in my testing, CN896 could only play materials at 1440x810 (from http://anon.nasa-global.edgesuite.net/anon.nasa-global/NASAHD/Lunar_Surface/LunarSurface_HD_1080i.wmv , reencoded to MPEG-2 with MEncoder at nearly the same bit rate) with a little lag. And just as described in VeMP's documentation, occasionally the machine died while playing. From Beijing, China ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to [EMAIL PROTECTED]
