We recently took delivery of some Intel motherboards that had IDE sockets but used the Marvell chipset. This uses a pata driver and infortunately these drivers are not put into the initramfs - thus boot fails with a 'no boot device found' message. The following procedure was done on a 2.6.20 hardened r2 kernel. It works but some things I do not understand and someone more knowledgeable should check this. Edit the following file: /usr/share/genkernel/x86/modules_load Add the line MODULES_PATA="pata_mpiix pata_pdc2027x pata_radisys pata_sil680 pata_rz1000 pata_efar pata_cmd64x pata_hpt366 pata_hpt37x pata_hpt3x3 pata_optidma pata_it821x pata_artop pata_oldpiix pata_cypress pata_platform pata_serverworks pata_legacy pata_ns87410 pata_pcmcia pata_isapnp pata_ali pata_amd pata_opti pata_atiixp pata_triflex pata_pdc202xx_old pata_sc1200 pata_qdi pata_netcell pata_sis pata_hpt3x2n pata_marvell pata_jmicron pata_via pata_cs5520 pata_cs5530 pata_cs5535 pata_sl82c105" I have added every pata module except pata_winbond which gave an error on boot. Then edit the file /usr/share/genkernel/generic/initrd.defaults. Change the entries for HWOPTS and add the pata group HWOPTS='usb firewire keymap cache evms2 pata sata lvm2 dmraid slowusb fs' MY_HWOPTS='usb firewire pata sata dmraid fs net' Rebuild your kernel or initramfs by executing genkernel initrd This kernel/initramfs will now boot off a CD on PATA motherboard. Some questions Do I have to change both HWOPTS and MY_HWOPTS in initrd.defaults ? The genkernel script is quite capable of generating MY_HWOPTS dynamically and this would be an improvement. At the top of the genkernel script, the modules_load file is source'd but the loop following this source statement references ${!AMODULES_*} instead of ${!MODULES_*}. Is this an error ? [ genkernel 3.4.6 ] There is an append_modules() function in gen_initramfs.sh that does not appear to be used anywhere ? Reproducible: Always Steps to Reproduce: 1.Boot off live CD on PATA motherboard 2. 3. Actual Results: See description above Expected Results: Successful boot off live CD on pata motherboard emerge --info Portage 2.1.2.7 (hardened/x86/2.6, gcc-3.4.6, glibc-2.3.6-r5, 2.6.20-hardened-r2 i686) ================================================================= System uname: 2.6.20-hardened-r2 i686 Intel(R) Celeron(R) CPU 2.40GHz Gentoo Base System release 1.12.9 Timestamp of tree: Mon, 21 May 2007 09:00:01 +0000 ccache version 2.4 [enabled] dev-java/java-config: 1.3.7, 2.0.31-r5 dev-lang/python: 2.4.4-r4 dev-python/pycrypto: 2.0.1-r5 dev-util/ccache: 2.4-r7 sys-apps/sandbox: 1.2.17 sys-devel/autoconf: 2.60 sys-devel/automake: 1.4_p6, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2, 1.10 sys-devel/binutils: 2.16.1-r3 sys-devel/gcc-config: 1.3.16 sys-devel/libtool: 1.5.22 virtual/os-headers: 2.6.17-r2 ACCEPT_KEYWORDS="x86" AUTOCLEAN="yes" CBUILD="i686-pc-linux-gnu" CFLAGS="-mtune=i686 -Os -pipe -fforce-addr" CHOST="i686-pc-linux-gnu" CONFIG_PROTECT="/etc /usr/lib/fax /var/spool/fax/etc" CONFIG_PROTECT_MASK="/etc/env.d /etc/env.d/java/ /etc/gconf /etc/java-config/vms/ /etc/php/apache2-php5/ext-active/ /etc/php/cgi-php5/ext-active/ /etc/php/cli-php5/ext-active/ /etc/revdep-rebuild /etc/terminfo" CXXFLAGS="-mtune=i686 -Os -pipe -fforce-addr" DISTDIR="/usr/portage/distfiles" FEATURES="ccache distlocks metadata-transfer parallel-fetch sandbox sfperms strict" GENTOO_MIRRORS="http://distfiles.gentoo.org http://distro.ibiblio.org/pub/linux/distributions/gentoo" MAKEOPTS="-j2" PKGDIR="/usr/src/asterisk" PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress --force --whole-file --delete --delete-after --stats --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages --filter=H_**/files/digest-*" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" SYNC="rsync://rsyncproxy/gentoo-portage" USE="alsa berkdb bluetooth cdr cgi cli crypt dba dbm dvd hal hardened kerberos ldap ldapsam midi netboot nls pam pic ppds readline samba sqlite3 srvdir ssl tcpd urandom x86 xorg zlib" ALSA_PCM_PLUGINS="adpcm alaw asym copy dmix dshare dsnoop empty extplug file hooks iec958 ioplug ladspa lfloat linear meter mulaw multi null plug rate route share shm softvol" ELIBC="glibc" INPUT_DEVICES="mouse keyboard" KERNEL="linux" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses text" USERLAND="GNU" Unset: CTARGET, EMERGE_DEFAULT_OPTS, INSTALL_MASK, LANG, LC_ALL, LDFLAGS, LINGUAS, PORTAGE_COMPRESS, PORTAGE_COMPRESS_FLAGS, PORTAGE_RSYNC_EXTRA_OPTS, PORTDIR_OVERLAY
Please try w/ genkernel-3.4.8; also, which module you need added? There's no problem w/ booting from PATA here.
(In reply to comment #1) > Please try w/ genkernel-3.4.8; also, which module you need added? There's no > problem w/ booting from PATA here. > I have tried 3.4.8 but no change. I still had to add the MODULES_PATA variable to /usr/share/genkernel/x86/modules_load. The motherboard is an Intel DG965wh. Incidentally the Gentoo 2007.0 minimal install CD will not boot this M/B. I booted a SuSe 10.2 and noted that it loaded the pata_marvell module. Adding this module enables booting from a Live Cd on this M/B. There are BIOS options to make the disk interface 'Legacy' but this has no effect - you must load the pata_marvell module before you can boot and this means you must install this module in the initramfs and that is why you need to change modules_load and initrd.defaults.
Reopen.
The reason it's not loaded automatically is because we don't use the new PATA drivers in the default genkernel configs. Why are you using genkernel to build a custom kernel, anyway?
(In reply to comment #4) > The reason it's not loaded automatically is because we don't use the new PATA > drivers in the default genkernel configs. Why are you using genkernel to build > a custom kernel, anyway? > We're not building a custom kernel. We have a master PC from which we ghost our production machines (which can be any M/B). So our kernel must be generic similarly to the Gentoo 2007/0 minimal install CD or any other live CD (knoppix etc.). Other live CD's of recent vintage boot off this M/B except Gentoo simply because Gentoo does not (as you say) load the PATA drivers. Note that this bug does *not* just pertain to our ghosting process - the standard Gentoo Live CD/Install CD's also do not boot off this M/B and the only fix is to load the PATA drivers from initramfs. The Suse 10.2 DVD does boot - but then it does auto-detect the pata_marvell driver that this M/B requires. The Gentoo initramfs seems not to do auto-detection but blindly loads all drivers that it scans for i.e those listed in /usr/share/genkernel/x86/modules_load
Any progress ? I am willing to test any changes - I have 2 of the offending motherboards
I have just tried Startcom Enterprise 5 Linux which is a free version of RHEL5. It does not boot off the CD either. One can boot the kernel - it loads a driver and then cannot autodetect the boot device (much as Gentoo also fails to do) - a dialog is then displayed and you can select a driver - unfortunately the pata_marvell selection is unavailable. I tried setting the BIOS to Native-IDE or Native-AHCI but neither worked. The closest driver selection available is the sata_mv driver. Fedora 6 fails identically to Startcom. Ubuntu 6.1 fails dramatically - you are dropped into an ash shell and the log on console 1 shows that it was unable to access /root/ /proc and other filesystems Windows 2000, Windows XP SP2 both boot successfully and will format the disk and start copying files.
Annoying us with "is it done yet?" and useless information won't get us to do anything faster. You've already told us how to fix it. We'll get around to it when we get around to it.
(In reply to comment #0) > Do I have to change both HWOPTS and MY_HWOPTS in initrd.defaults ? Yes. HWOPTS is a list of all options, MY_HWOPTS is a list of what is enabled. > The genkernel script is quite capable of generating MY_HWOPTS dynamically > and this would be an improvement. Feel free to file a bug with that patch and I'll add it, but MY_HWOPTS changes depending on the command line options passed. > At the top of the genkernel script, the modules_load file is source'd but > the loop following this source statement references ${!AMODULES_*} instead of > ${!MODULES_*}. Is this an error ? [ genkernel 3.4.6 ] Nope. It's done on purpose for different functionality. > There is an append_modules() function in gen_initramfs.sh that does not > appear to be used anywhere ? It should be used by the code that also uses AMODULES_* but it definitely is not being used. I'm not really sure if there's a bug in some other code and it *should* be getting called and isn't or if it is just useless now and should be removed. Anyway, I would like to thank you for the code below and I have added it to genkernel SVN. It'll show up in the next genkernel version. What error did you get with winbond, do you remember?
(In reply to comment #9) > (In reply to comment #0) The genkernel script is quite capable of generating MY_HWOPTS dynamica> > > > Feel free to file a bug with that patch and I'll add it, but MY_HWOPTS changes > depending on the command line options passed. No I will leave this alone > > Anyway, I would like to thank you for the code below and I have added it to > genkernel SVN. It'll show up in the next genkernel version. What error did > you get with winbond, do you remember? > Thanks. Unfortunately I cannot remember the winbond error. I tried to repliacte it but we have already remade all our LiveCD's without winbond. I will be building machines late next week and will force the error then and report back if I can. When will your changes find their way into mainstream? When they appear I will try them on these M/B
If you have the time to replicate the error, I would love to see it. If not, I'll just take your word for it and try to figure out where to go from there for the winbond driver. The first step would likely be to figure out how popular the chipset is in the wild and then decide if we can afford to not support it or not. As for the new genkernel, I am actually planning on making the next release genkernel 3.5, which will include a few more changes than a 3.4.9 release. If you're willing to test an unreleased version of genkernel, I can roll one up especially for you to try out this functionality. I'd definitely like to know the new PATA support works before I roll 3.5 out. I just don't see myself getting all the bugs fixed and new functionality added to genkernel in a speedy manner and I don't want to miss the opportunity to have someone with the hardware test it. Of course, you can roll your own, too. All I am going to do is package up what I have in SVN now as a tarball. It should be functional as-is from SVN right now. If not, I'd like to get *that* fixed, too... :P
For the record, please note that I tried to boot off these motherboards using the 'doload' option from the grub boot prompt i.e. typing 'gentoo doload=pata_marvell' and this did not work with *any* combination of settings in the bias. (In reply to comment #11) > If you have the time to replicate the error, I would love to see it. If not, > I'll just take your word for it and try to figure out where to go from there > for the winbond driver. The first step would likely be to figure out how > popular the chipset is in the wild and then decide if we can afford to not > support it or not. > I rebuilt my initramfs with pata_winbond added back and rebooted my PC (which is *not* one of the offending motherboards) and the error from 'dmesg' is (repeated many times) : x0 irq 14 IRQ handler type mismatch for IRQ 14 current handler: ide0 [<c0183651>] [<f88d737e>] [<f88d305c>] [<c01837e0>] [<f88d3b05>] [<f88db425>] [<f88da5f8>] [<c02af7a1>] [<f8d4019d>] [<c036b2b6>] [<c0163ee3>] [<f8d40232>] [<f8952284>] [<c01799d2>] [<c0145cf0>] ======================= platform pata_winbond.0: irq 14 request failed: -16 ata1027: PATA max PIO4 cmd 0x1F0 ctl 0x3F6 bmdma 0x0 irq 14 IRQ handler type mismatch for IRQ 14 current handler: ide0 [<c0183651>] [<f88d737e>] [<f88d305c>] [<c01837e0>] [<f88d3b05>] [<f88db425>] [<f88da5f8>] [<c02af7a1>] [<f8d4019d>] [<c036b2b6>] [<c0163ee3>] [<f8d40232>] [<f8952284>] [<c01799d2>] [<c0145cf0>] [<c0144d3b>] [<c0145d11>] [<c0360033>] ======================= platform pata_winbond.0: irq 14 request failed: -16 ata1028: PATA max PIO4 cmd 0x1F0 ctl 0x3F6 bmdma 0x0 irq 14 IRQ handler type mismatch for IRQ 14 current handler: ide0 [<c0183651>] [<f88d737e>] [<f88d305c>] [<c01837e0>] [<f88d3b05>] [<f88db425>] [<f88da5f8>] [<c02af7a1>] [<f8d4019d>] [<c036b2b6>] [<c0163ee3>] [<f8d40232>] [<f8952284>] [<c01799d2>] [<c0145cf0>] [<c014007b>] [<c0144d3b>] [<c0145d11>] ======================= platform pata_winbond.0: irq 14 request failed: -16 ata1029: PATA max PIO4 cmd 0x1F0 ctl 0x3F6 bmdma 0x0 irq 14 IRQ handler type mismatch for IRQ 14 current handler: ide0 [<c0183651>] [<f88d737e>] [<f88d305c>] [<c01837e0>] [<f88d3b05>] [<f88db425>] [<f88da5f8>] [<c02af7a1>] [<f8d4019d>] [<c036b2b6>] [<c0163ee3>] [<f8d40232>] [<f8952284>] [<c01799d2>] [<c0145cf0>] ======================= platform pata_winbond.0: irq 14 request failed: -16 ata1030: PATA max PIO4 cmd 0x1F0 ctl 0x3F6 bmdma 0x0 irq 14 IRQ handler type mismatch for IRQ 14 current handler: ide0 [<c0183651>] [<f88d737e>] [<f88d305c>] [<c01837e0>] [<f88d3b05>] [<f88db425>] [<f88da5f8>] [<c02af7a1>] [<f8d4019d>] [<c036b2b6>] [<c0163ee3>] [<f8d40232>] [<f8952284>] [<c01799d2>] [<c0145cf0>] [<c01a007b>] ======================= platform pata_winbond.0: irq 14 request failed: -16 ata1031: PATA max PIO4 cmd 0x1F0 ctl 0x3F6 bmdma 0x0 irq 14 IRQ handler type mismatch for IRQ 14 current handler: ide0 [<c0183651>] [<f88d737e>] [<f88d305c>] [<c01837e0>] [<f88d3b05>] [<f88db425>] [<f88da5f8>] [<c02af7a1>] [<f8d4019d>] [<c036b2b6>] [<c0163ee3>] [<f8d40232>] [<f8952284>] [<c01799d2>] [<c0145cf0>] [<c036007b>] ======================= platform pata_winbond.0: irq 14 request failed: -16 ata1032: PATA max PIO4 cmd 0x1F0 ctl 0x3F6 bmdma 0x0 irq 14 IRQ handler type mismatch for IRQ 14 current handler: ide0 [<c0183651>] [<f88d737e>] [<f88d305c>] [<c01837e0>] [<f88d3b05>] [<f88db425>] [<f88da5f8>] [<c02af7a1>] [<f8d4019d>] [<c036b2b6>] [<c0163ee3>] [<f8d40232>] [<f8952284>] [<c01799d2>] [<c0145cf0>] ======================= etc etc.... With the changes I originally did (from my initial post), all the pata drivers are loaded and sit in memory. I can load and unload them using the modprobe command without error. The reloading of the module after unloading it is slow. Only the pata_winbond module reports 'no such device' and generates the above errors in the dmesg log. I would speculate that the winbond driver is not coded to the same standard as the others. > you're willing to test an unreleased version of genkernel, I can roll one up > especially for you to try out this functionality. I'd definitely like to know > the new PATA support works before I roll 3.5 out. I just don't see myself > getting all the bugs fixed and new functionality added to genkernel in a speedy > manner and I don't want to miss the opportunity to have someone with the > hardware test it. Of course, you can roll your own, too. All I am going to do > is package up what I have in SVN now as a tarball. It should be functional > as-is from SVN right now. If not, I'd like to get *that* fixed, too... :P > I will test it - how will you send it to me ? (patch ?)
Actually, I've got quite a few people that need to test things, so I'll be rolling out a genkernel-3.4.9_pre1 shortly.
Please test genkernel 3.4.9_prer1 or better. This should be fixed now.
