Hello Peter & Jeremy, I've found an ordinary desktop PC (with Phoenix Award BIOS 6.00PG) that won't boot off a USB stick created by livecd-tools-024 with syslinux (tested both versions 3.75 and 3.81). The boot process drops to the "boot:" prompt with an error message: could not find kernel image: linux The same USB stick boots fine on any other computer I could find. Does it seem like a syslinux bug? And if turns out to be a known BIOS bug, is there a good workaround? -- // Bernie Innocenti - http://codewiz.org/ \X/ Sugar Labs - http://sugarlabs.org/
Bernie Innocenti wrote:> Hello Peter & Jeremy, > > I've found an ordinary desktop PC (with Phoenix Award BIOS 6.00PG) that > won't boot off a USB stick created by livecd-tools-024 with syslinux > (tested both versions 3.75 and 3.81). > > The boot process drops to the "boot:" prompt with an error message: > > could not find kernel image: linux > > The same USB stick boots fine on any other computer I could find. > Does it seem like a syslinux bug? And if turns out to be a known BIOS > bug, is there a good workaround? >I need much more details; *all* Award BIOSes make in the past 10-12 years have version number 6.00PG. Also look for how you have configured your BIOS... some Award BIOSes have USB-ZIP, USB-HDD, USB-FDD configurations; you generally want USB-HDD. There are some BIOSes which will boot from USB *only* if it was formatted with 64 heads, 32 sectors; apparently due to some odd notion that only zipdrives would be USB. -hpa -- H. Peter Anvin, Intel Open Source Technology Center I work for Intel. I don't speak on their behalf.
On Monday, June 1, 2009, 7:32:56, Bernie Innocenti wrote:> The same USB stick boots fine on any other computer I could find. > Does it seem like a syslinux bug? And if turns out to be a known BIOS > bug, is there a good workaround?It's most likely a BIOS bug. You'll probably find out that syslinux prints CBIOS in it's welcome line, meaning the BIOS is limiting it to drives < 1024 cylinders. If your USB stick is formatted with ZIP drive geometry, try formatting it normally - at least for me, that solved the booting problem on Gigabyte G33-DS3R motherboard. Another thing you could try is that instead of booting the stick as USB-HDD (or USB-ZIP), press the key to display boot menu at bootup, choose +Hard drive there, which should give you another menu, and select your USB stick from that menu. -- < Jernej Simon?i? ><><><><>< http://eternallybored.org/ > Don't ask the barber whether you need a haircut. -- First Law of Expert Advice
Luke and Sasha are working on a new USB format that they feel will allow more machines to boot and support VM + Stick http://wiki.sugarlabs.org/go/Sugar_on_a_Stick/USB_format Perhaps the issue you are running into (which we have definitely seen before) is related to some of the ones they are looking at and part of why they need two boot partitions that are slightly different. On Mon, Jun 1, 2009 at 1:32 AM, Bernie Innocenti <bernie at codewiz.org> wrote:> Hello Peter & Jeremy, > > I've found an ordinary desktop PC (with Phoenix Award BIOS 6.00PG) that > won't boot off a USB stick created by livecd-tools-024 with syslinux > (tested both versions 3.75 and 3.81). > > The boot process drops to the "boot:" prompt with an error message: > > could not find kernel image: linux > > The same USB stick boots fine on any other computer I could find. > Does it seem like a syslinux bug? And if turns out to be a known BIOS > bug, is there a good workaround? > > -- > // Bernie Innocenti - http://codewiz.org/ > \X/ Sugar Labs - http://sugarlabs.org/ > _______________________________________________ > Sugar-devel mailing list > Sugar-devel at lists.sugarlabs.org > http://lists.sugarlabs.org/listinfo/sugar-devel >-- Caroline Meeks Solution Grove Caroline at SolutionGrove.com 617-500-3488 - Office 505-213-3268 - Fax
Caroline Meeks wrote:> Luke and Sasha are working on a new USB format that they feel will allow > more machines to boot and support VM + Stick > > http://wiki.sugarlabs.org/go/Sugar_on_a_Stick/USB_format > > Perhaps the issue you are running into (which we have definitely seen > before) is related to some of the ones they are looking at and part of > why they need two boot partitions that are slightly different. >I don't understand why 128 heads. 64 heads is the more compatible version. I presume the notion of partition 1 and 4 is to deal with things that have an odd notion of zipdisks. I have personally not seen any devices which will only boot with partition 1 or only with partition 4, if you have any such information I would appreciate it. Part of me also wonders if using EXTLINUX might not be easier for you, too. -hpa -- H. Peter Anvin, Intel Open Source Technology Center I work for Intel. I don't speak on their behalf.
On Mon, Jun 01, 2009 at 11:38:05AM -0400, Caroline Meeks wrote:>> I've found an ordinary desktop PC (with Phoenix Award BIOS 6.00PG) >> that >> won't boot off a USB stick created by livecd-tools-024 with syslinux >> (tested both versions 3.75 and 3.81). >> >> The boot process drops to the "boot:" prompt with an error message: >> >> could not find kernel image: linuxIf it already gets to this stage, the new partition setup won't help.>> The same USB stick boots fine on any other computer I could find.Very strange and unfortunately out of my area of knowledge. One last idea: Have you tried other USB sticks yet? Perhaps it's a hardware incompatibility.>> Does it seem like a syslinux bug? And if turns out to be a known >> BIOS >> bug, is there a good workaround?FWIW, the machine (Tyan S2495 based) I'm doing some USB stick testing on also claims to be Phone Award BIOS 6.00PG and it works fine (albeit slow). CU Sascha -- http://sascha.silbe.org/ http://www.infra-silbe.de/ -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 489 bytes Desc: Digital signature URL: <http://www.zytor.com/pipermail/syslinux/attachments/20090601/bd554d4c/attachment.sig>
On 06/01/09 08:39, H. Peter Anvin wrote:> I need much more details; *all* Award BIOSes make in the past 10-12 > years have version number 6.00PG.Ouch, I no longer have access to it. I asked the owner to let me know.> Also look for how you have configured your BIOS... some Award BIOSes > have USB-ZIP, USB-HDD, USB-FDD configurations; you generally want USB-HDD.I booted from the BBS menu (F8), the item was labelled something like "USB HDD 2.0".> There are some BIOSes which will boot from USB *only* if it was > formatted with 64 heads, 32 sectors; apparently due to some odd notion > that only zipdrives would be USB.Good to know. Mine looks like this: bernie at giskard:~/src/kernel$ sudo fdisk -l /dev/sdb Disk /dev/sdb: 2055 MB, 2055208960 bytes 221 heads, 2 sectors/track, 9081 cylinders Units = cylinders of 442 * 512 = 226304 bytes Disk identifier: 0x000aa8c2 Device Boot Start End Blocks Id System /dev/sdb1 * 1 9082 2007006+ c W95 FAT32 (LBA) -- // Bernie Innocenti - http://codewiz.org/ \X/ Sugar Labs - http://sugarlabs.org/
-----BEGIN PGP SIGNED MESSAGE----- Hash: RIPEMD160 On Tue, Jun 02, 2009 at 02:40:38AM +0200, Bernie Innocenti wrote:>On 06/01/09 08:39, H. Peter Anvin wrote: >> I need much more details; *all* Award BIOSes make in the past 10-12 >> years have version number 6.00PG. > >Ouch, I no longer have access to it. I asked the owner to let me know. > > >> Also look for how you have configured your BIOS... some Award BIOSes >> have USB-ZIP, USB-HDD, USB-FDD configurations; you generally want USB-HDD.In my (older non-Sugar) experience, USB-HDD is best, then USB-ZIP, and if none of those options are available then pick USB-FDD (which is then most likely names something else). Whatever you pick, make sure to mention the choice when reporting trouble, to help recognize similarities in use cases.>I booted from the BBS menu (F8), the item was labelled something like >"USB HDD 2.0".Another important thing (not sure if it has been mentioned before) is to *cold* boot with stick already inserted. That is, insert the USB stick before turning on the machine - and after using BIOS or some operating system then power down the machine completely, wait a few seconds, and then boot it. The reason for this (as I understand it) is that the simple BIOS USB drivers often do not reliably handle switching state. This also means that a last desperate test could be to 1) boot that other operating system (Windows or Mac), 2) insert USB stick and 3) reboot (*without* shutting down completely). It might work, but you then have the heavy and slow burden of your USB boot routine including a full bootup and shutdown of another operating system. :-/ - Jonas - -- * Jonas Smedegaard - idealist og Internet-arkitekt * Tlf.: +45 40843136 Website: http://dr.jones.dk/ [x] quote me freely [ ] ask before reusing [ ] keep private -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) iEYEAREDAAYFAkok1DEACgkQn7DbMsAkQLjhoQCfeSfnmSHJ6CTQ/twxIMrXAlZl KqoAn0XUHRxuk+sBYzmidKzjU9JyhbOw =avFw -----END PGP SIGNATURE-----
> > On a side note, which filesystem should be chosen in order to minimize > wearing to USB sticks? >My vote would be squashfs.
On Sun, Aug 30, 2009 at 21:37, Mitch Bradley <wmb at firmworks.com> wrote:> >> On a side note, which filesystem should be chosen in order to minimize >> wearing to USB sticks? >> > My vote would be squashfs.What about for the writable part of the filesystem, since squashfs is read-only? -- Luke Faraone http://luke.faraone.cc
> > On Sun, Aug 30, 2009 at 21:37, Mitch Bradley <wmb at firmworks.com > <mailto:wmb at firmworks.com>> wrote: > > > On a side note, which filesystem should be chosen in order to > minimize > wearing to USB sticks? > > My vote would be squashfs. > > > What about for the writable part of the filesystem, since squashfs is > read-only?That was meant to be a joke, albeit one with a point lurking inside it. The only way to truly minimize wear is not to write. On the topic of FAT, one might imagine a vendor optimizing for the consumer use case. Consumers almost always use the device as-formatted from the factory. If you know that the device has a FAT filesystem with a known layout, you could optimize the FTL for that case. You know that the FAT is a hot spot, but you know where the FAT is located, so you do special things to prevent wear at that virtual location.
On Sun, Aug 30, 2009 at 09:33:22PM -0400, Bernie Innocenti wrote:>El Sun, 30-08-2009 a las 20:12 +0200, Jonas Smedegaard escribi?: > >> Yes, the name of the tool is misleading: It works with other fs types >> too - its main purpose it partitioning, not fs formatting. I >> succesfully booted USB sticks formatted as ext2 (as far as I recall - >> I last worked with it a year ago). > >On a side note, which filesystem should be chosen in order to minimize >wearing to USB sticks? > >I would assume that the any DOS filesystem will continouously rewrite >to the FAT blocks. Maybe the best choice would be ext4 with the >journal disabled?For read-only parts: some compressed filesystem (to minimize read wear), e.g. squashfs or the FUSE2-based fusecompress write-protected. For read/write parts: some filesystem without journaling, best a log fs (i.e. "rotating" fs), e.g. nilfs2. ...but that is looking only at wear. Taking reliability and performance into account too, I would use ext4 without journaling (possibly even for read-only stuff too, just write-protected, to ease administration and possibly also memory concumption and performance due to simpler and shared cache handling!). A hint about ext4 disabling journaling: There is 2 ways of doing it - setting a flag in the ext4 metadata and setting a flag in the root dir metadata. The first is the obvious one, but was introduced only in Linux 2.6.30 (or was it 2.6.29?). The other method works also with earlier incarnations of ext4, but works only if you set it as the very first activity on the filesystem, as newly created objects inherit features from the dir they were created in. - Jonas -- * Jonas Smedegaard - idealist & Internet-arkitekt * Tlf.: +45 40843136 Website: http://dr.jones.dk/ [x] quote me freely [ ] ask before reusing [ ] keep private -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 835 bytes Desc: Digital signature URL: <http://www.zytor.com/pipermail/syslinux/attachments/20090831/3e1fe4ff/attachment.sig>