Thanks for the reply, I used Flashrom as the utility and am quite comfortable using Linux and the command line. I got shot of my virally infected copy of Win98 sometime ago, I have my hard drive with Fedora 9 on my friends PC, it has the Flashrom utility and my backup.bin. It also has the same type of bios chip (PLCC), And its socketed!
My bios chip is a PLCC (not socketed unfortunately) and the info on the chip is;
Winbond W39V040AP 240515401 415GGAA
I've thought about unsoldering the chip but don't really have the proper tools. I could do it though, it will take ages because I'll have to take it real slow so and find something to use to take the heat away from the chip so I don't fry it, but I'm very patient and I think could do it (getting it back on might be a bit more problematic).
My thought was to purchase another chip and 'kind of' place it on top of the other chip, but I'm not sure how this would go? I'd hope it would be a little like hot flashing, and the good chip would be recognized by the CPU and I could boot-up. It would then be a case of removing the new chip (making sure the bios is set to cashable first) and refashing with my backup.bin. But I don't know the 'bad' chip could interfere with it booting and cause it not to work. Another thought is this computer has something in the bios about 'booting' 'rom' and 'lan' I haven't fully looked into it yet, but if its what I think it is, its something to do with booting over a network, is it possible to do this with a CPU that has no bios?
I'm open and confident in trying all suggestions, and I love to learn.
Thank you
--- On Thu, 11/6/09, Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net wrote:
From: Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net Subject: Re: [coreboot] Hosed ms-7032, I foolishly flashed the bios with a .700 To: "Phil Neary" philneary@yahoo.co.uk Cc: coreboot@coreboot.org Date: Thursday, 11 June, 2009, 10:20 AM Hi Phil,
On 11.06.2009 11:49, Phil Neary wrote:
Hi all, I'm not really sure how this works, I'm
guessing its a bit similar to a forum, but in lieu of creating threads and posting posts, you email instead?
Correct. Please make sure to use the "reply to all" button in your mailer. That will ensure the individual developers and the list see your mails.
[...] I messed up when flashing my bios and flashed it
with A7032VMS.700 instead of a .rom or a .bin.
Which application did you use for reflashing? In theory, the A7032VMS.700 may just be a renamed .bin file.
[...] I tried the AMIBOOT.ROM recovery with the .700
and a few other bios' for my board [...], but they haven't worked.[...] I was able to get the backup.bin from my hard drive
Actually, the backup.bin may be key to reviving your board.
and tried renaming it AMIBOOT.ROM and saving it to
floppy, but Fedora named it amiboot.rom[...]. Does Amiboot have to be capitalized?
AFAIK DOS doesn't care about name capitalization.
Is there anything I haven't tried yet?
It all depends on how adventurous you feel. Can you try to locate the ROM chip on your board and peel off the sticker? It might look similar to this: http://www.coreboot.org/File:Plcc32_chip.jpg and it is probably in a socket. Write down all text you see on the chip (not the sticker) and mail it to us. Then we can tell you which chips are compatible and how to recover. One way to recover is to hotflash the chip in another board with compatible flash bus (sounds complicated, but once we know the chip model, we can tell you which boards are compatible). That would save you the hassle of buying a new chip.
I can get a replacement chip for a fiver, but before I
do I'd like to find out if that will defiantly work?
It should work, but it's not guaranteed. Ah yes, and there is the risk that your onboard network won't work anymore if you buy a preflashed chip because sometimes the old chip stores the MAC address of your onboard network card.
How experienced are you with Linux or *BSD? Our emergency recovery tools don't work under Windows (well, very old versions of our tools work under Windows).
Regards, Carl-Daniel
Hi Phil,
On 11.06.2009 14:08, Phil Neary wrote:
I used Flashrom as the utility and am quite comfortable using Linux and the command line.
Good.
[..] my friends PC, it has the Flashrom utility and my backup.bin. It also has the same type of bios chip (PLCC), And its socketed!
Even better.
My bios chip is a PLCC (not socketed unfortunately) and the info on the chip is;
Winbond W39V040AP
Ah, the Winbond W39 series. I reread the data sheets dozens of times and I never found out why these chips sometimes commit suicide during flashing. It works for some mainboards and fails horribly for others.
I've thought about unsoldering the chip but don't really have the proper tools.
Some people on this list have experience with desoldering chips. Maybe they can assist.
My thought was to purchase another chip and 'kind of' place it on top of the other chip, but I'm not sure how this would go?
Yes, this would work with a bit of soldering. I looked through the data sheets and came up with a way to do this: Desolder the ID2 pin of the old chip from the board and connect it to VDD (3.3V). Solder the new chip (with the correct BIOS image) on top of the old chip, but connect ID2 of that chip to ground. Once that is done, the machine should boot normally. After booting, you'll have to modify flashrom to target the old chip or you'll simply overwrite the new chip which would be rather undesiarable.
Please try to avoid Winbond chips if you buy a replacement. The technical specs of the chip you want are: 4 Mbit LPC. The form factor of the chip is probably PLCC32, but you should verify that. I've good experiences with ST and SST chips.
It would then be a case of removing the new chip
Not entirely. Desoldering a chip in a running board is not exactly recommendable. Fortunately, with the ID2 trick, you can access two chips at once.
(making sure the bios is set to cashable first) and reflashing with my backup.bin.
Please make sure the BIOS is _NOT_ set to cacheable. Caching in the BIOS area is a sure-fire way to write garbage to a chip during flashing.
But I don't know the 'bad' chip could interfere with it booting
With the ID2 trick, this should work fine.
[...] something in the bios about 'booting' 'rom' and 'lan' [...] booting over a network, is it possible to do this with a CPU that has no bios?
Unfortunately not. Basically, when you switch on your computer, RAM does not work. The BIOS (or coreboot) takes care of switching on RAM. Without that, booting over the network is impossible.
May I suggest you drop by on IRC in #coreboot at irc.freenode.net. Most of us are present in that chatroom and you might get faster answers (no guarantee about answer speed, though). It will help if you mention there that you already sent a mail with details to the coreboot list.
Regards, Carl-Daniel
On Thu, Jun 11, 2009 at 02:51:41PM +0200, Carl-Daniel Hailfinger wrote:
I've thought about unsoldering the chip but don't really have the proper tools.
Some people on this list have experience with desoldering chips. Maybe they can assist.
If you don't mind sacrificing the current chip, you could just cut it off and replace it with a socket. See
http://www.coreboot.org/Soldering_a_socket_on_your_board#Cutting_the_chip
and for more detail (shameless plug):
http://ward.vandewege.net/blog/2009/03/howto-replace-a-plcc-chip-with-a-sock...
It's not hard to do, even with moderate soldering skills.
Thanks, Ward.
On Thu, Jun 11, 2009 at 23:10, Ward Vandewegeward@gnu.org wrote:
If you don't mind sacrificing the current chip, you could just cut it off and replace it with a socket. See
http://www.coreboot.org/Soldering_a_socket_on_your_board#Cutting_the_chip
Another one I've used in the past, if you're REALLY careful about it you can use an xacto knife or similar to "liberate" the chip from the board, essentially you push the tip of the knife into junction of the solder and the leg you can usually "cut" the solder, solder is relatively soft in comparison to the leg so you're usually OK on that front (and even if you do sever the leg towards the bottom chances are you can still use the chip in a socket) however there is more scope to damage the tracks/pads on the mainboard (That said, I'm not sure whether lead-free solder is as easy to "cut" as leaded solder), I used to use this trick to safely extract RAM chips off RAM sticks to make a double sided module out two single sided ones ;oD
Ward Vandewege wrote:
and for more detail (shameless plug):
http://ward.vandewege.net/blog/2009/03/howto-replace-a-plcc-chip-with-a-sock...
It's not hard to do, even with moderate soldering skills.
Let me introduce you to a fantastic product. We have been using it at OLPC with great success.
[you will have to overlook the amateur web page but the product is awesome]
The issue most non-experienced (and even experienced) people have with SMT part removal is ripping off the pads due to excessive heat and force.
Chipquik is an alloy that drops the melting point of solder to less than 150 degF. So literally you can remove a part with a hairdryer. Its wonderful stuff. All you have to do is make sure that you clean it up good so that when you solder the part back on you don't have any chipquik left behind.
On Thu, Jun 11, 2009 at 22:51, Carl-Daniel Hailfingerc-d.hailfinger.devel.2006@gmx.net wrote:
Yes, this would work with a bit of soldering. I looked through the data sheets and came up with a way to do this: Desolder the ID2 pin of the old chip from the board and connect it to VDD (3.3V). Solder the new chip (with the correct BIOS image) on top of the old chip, but connect ID2 of that chip to ground. Once that is done, the machine should boot normally. After booting, you'll have to modify flashrom to target the old chip or you'll simply overwrite the new chip which would be rather undesiarable.
I've just done some reading, that should do the trick. Except according to the datasheet you should use ID3 for this not ID2, essentially the ID pins control where in the 8M of space the contents of the chip gets mapped, by using ID3 you can flip your 4M ROM between the top and the bottom of the address range, as I say I've not looked too deeply into LPC but my reading of the datasheet suggests that if you use ID2 you will end up with half of your two ROMs overlapping.
On 11.06.2009 15:12, Morgan Reed wrote:
On Thu, Jun 11, 2009 at 22:51, Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net wrote:
Desolder the ID2 pin of the old chip from the board and connect it to VDD (3.3V). Solder the new chip (with the correct BIOS image) on top of the old chip, but connect ID2 of that chip to ground.
I've just done some reading, that should do the trick. Except according to the datasheet you should use ID3 for this not ID2, essentially the ID pins control where in the 8M of space the contents of the chip gets mapped, by using ID3 you can flip your 4M ROM between the top and the bottom of the address range, as I say I've not looked too deeply into LPC but my reading of the datasheet suggests that if you use ID2 you will end up with half of your two ROMs overlapping.
Ah, the common trap of MByte vs. Mbit. The chips are 4 Mbit (512 kByte), so there won't be any overlap. Besides that, the Winbond datasheets explicitly mention that ID3 is not used by the chips unless I'm totally misreading them.
Regards, Carl-Daniel
On Thu, Jun 11, 2009 at 23:30, Carl-Daniel Hailfingerc-d.hailfinger.devel.2006@gmx.net wrote:
Ah, the common trap of MByte vs. Mbit. The chips are 4 Mbit (512 kByte), so there won't be any overlap.
No, I'm well aware of the distinction (embedded hardware designer) you are right though, I misread the datasheet (Saw M assumed Mb because it was a ROM, looking at it again they say M byte (presumably because a lot of people don't realise the difference between Mb and MB), teach me for posting to mailing lists at 22:00 ;oD )
Besides that, the Winbond datasheets explicitly mention that ID3 is not used by the chips unless I'm totally misreading them.
From my datasheet (Revision A2, dated 19/12/2002);
...For accessing the 4M byte BIOS storage space, the ID[2:0] pins are inverted in the ROM and are compared to address lines [21:19]. ID[3] can be used as like active low chip-select pin.
Ergo, tie ID3 high, chip effectively disabled, same outcome as using the ID2 pin as you say though.
On 11.06.2009 16:15, Morgan Reed wrote:
On Thu, Jun 11, 2009 at 23:30, Carl-Daniel Hailfingerc-d.hailfinger.devel.2006@gmx.net wrote:
Ah, the common trap of MByte vs. Mbit. The chips are 4 Mbit (512 kByte), so there won't be any overlap.
No, I'm well aware of the distinction (embedded hardware designer)
Neat. I didn't want to imply you didn't know about the distinction, but I sometime confuse units when I'm very tired. Seems I'm not the only one. ;-)
From my datasheet (Revision A2, dated 19/12/2002);
...For accessing the 4M byte BIOS storage space, the ID[2:0] pins are
Whoa. 4 M byte. Evil datasheet.
Ergo, tie ID3 high, chip effectively disabled, same outcome as using the ID2 pin as you say though.
True. My idea in using ID2 was to make the old chip available a bit lower in the address space so flashrom could access it while the other chip is stacked on top.
Regards, Carl-Daniel
*grumble* Repost CC to the list...
On Fri, Jun 12, 2009 at 00:21, Carl-Daniel Hailfingerc-d.hailfinger.devel.2006@gmx.net wrote:
Whoa. 4 M byte. Evil datasheet.
Yah
True. My idea in using ID2 was to make the old chip available a bit lower in the address space so flashrom could access it while the other chip is stacked on top.
Hmm, that's a good point, I didn't actually realise flashrom supported multiple chips on the one bus but it does make sense, that way he can reflash the old BIOS (assuming the chip isn't dead), restore the connection and all is good.
I'd still recommend replacing it with a socket though (less pain in the future). Or alternatively if he's feeling adventurous, make it permanent and fit a switch to tie one or the other high (they have internal pulldowns on the ID lines), lift "Write Enable" from the functional BIOS and you're guaranteed to always have a working BIOS available at the flick of a switch (I suppose you could call it a ghetto BIOS saviour) ;oD
Resending, CC to the list.
On Thu, Jun 11, 2009 at 22:08, Phil Nearyphilneary@yahoo.co.uk wrote:
My thought was to purchase another chip and 'kind of' place it on top of the other chip, but I'm not sure how this would go? I'd hope it would be a little like hot flashing, and the good chip would be recognized by the CPU and I could boot-up. It would then be a case of removing the new chip (making sure the bios is set to cashable first) and refashing with my backup.bin. But I don't know the 'bad' chip could interfere with it booting and cause it not to work. Another thought is this computer has something in the bios about 'booting' 'rom' and 'lan' I haven't fully looked into it yet, but if its what I think it is, its something to do with booting over a network, is it possible to do this with a CPU that has no bios?
I assume you're intending to use a chip programmed elsewhere with your backup image?
I doubt your suggestion would work, I don't know enough about LPC (your BIOS is an LPC type ROM as opposed to Parallel Flash or SPI) to say for sure but in the case of a parallel ROM you could do it if you lifted the "Chip Select" line off the board and connected that to the "top" chip you'd have a chance but I suspect in the case of an LPC you'd be better off desoldering the chip and replacing it with a freshly programmed one (or better yet, replace it with a PLCC socket to save you trouble in the future.)
-
Carl-Daniel Hailfinger -
Morgan Reed -
Phil Neary -
Richard Smith -
Ward Vandewege