I recently got myself a MSI Z68MA-G45 board. It's running with an Intel Z68 chipset, which has some support in flashrom and coreboot.
First, I ran flashrom with the system on the board and it was able to correctly read the chip (see attached log). I uploaded the read data [1]. The UEFI interface proposed to save the UEFI image on usb storage [2]: I was then able to compare it to the read data from flashrom. It appears that the UEFI image are the last 0x600000 bytes of the BIOS ROM image (from 0x200000 to 0x7FFFFF).
While searching for the BIOS chip on the board hardware, I saw a set of pins called "JSPI1", right next to the BIOS SPI chip [3]. After some research [4][5], I used this pin layout: [6].
The only SPI flasher I have is a Bus Pirate, that I used with flashrom at different spispeeds, but it never worked (it was only returning either 0xff or 0x00) (see attached logs). I used connectors from an USB on-board connector (9 pins) but had to take them off the plastic as the pins disposal didn't match the JSPI1 disposal (the pins are closer to each other than on USB on-board connectors). Fearing that the connectors would touch each other, I put adhesive tape around to isolate them [7]. As it still didn't work, I put some of the connectors of the bus pirate (I have a probe cable) directly on the pins, but it was only possible for 4 of them and used my hand-made cables for the rest [8]. Though, it still didn't work.
Any lead on this would be welcome. According to http://www.flashrom.org/ISP it can be because of cable length: are the buspirate probe cables too long for this? [9]
[1] http://download.paulk.fr/coreboot/msi-z68ma-g45/bios.rom [2] http://download.paulk.fr/coreboot/msi-z68ma-g45/E7676IMS.H10.rom [3] http://download.paulk.fr/coreboot/msi-z68ma-g45/photos/spi-chip-pins.jpg [4] http://soj.mesdiscussions.net/soj/Electronique/Autres/projet-adaptateur-hard... [5] http://forum-en.msi.com/index.php?topic=139099.0 [6] http://download.paulk.fr/coreboot/msi-z68ma-g45/photos/spi-pins.png [7] http://download.paulk.fr/coreboot/msi-z68ma-g45/photos/connectors.jpg [8] http://download.paulk.fr/coreboot/msi-z68ma-g45/photos/on-pins-connectors.jp... [9] http://download.paulk.fr/coreboot/msi-z68ma-g45/photos/buspirate.jpg
-- Paul Kocialkowski
On Mon, 2012-06-18 at 10:40 +0000, Paul Kocialkowski wrote:
I recently got myself a MSI Z68MA-G45 board. It's running with an Intel Z68 chipset, which has some support in flashrom and coreboot.
However the board itself is not in coreboot yet(I suppose you want to add support for it): ls src/mainboard/msi/ gives: Kconfig ms6119 ms6147 ms6156 ms6178 ms7135 ms7260 ms9185 ms9282 ms9652_fam10
First, I ran flashrom with the system on the board and it was able to correctly read the chip (see attached log). I uploaded the read data [1]. The UEFI interface proposed to save the UEFI image on usb storage [2]: I was then able to compare it to the read data from flashrom. It appears that the UEFI image are the last 0x600000 bytes of the BIOS ROM image (from 0x200000 to 0x7FFFFF).
Did you power off the mainboard? if not maybe that's why the buspirate doesn't work.
While searching for the BIOS chip on the board hardware, I saw a set of pins called "JSPI1", right next to the BIOS SPI chip [3].
So your BIOS chip can be accessed trough both SPI and FWH?
The only SPI flasher I have is a Bus Pirate,
do you also have an Arduino uno? or an openmoko debug board left(if not I can give you one back without the openmoko ribbon cable this summer)?
Note that the openmoko debug board is very undocumented.
Also Note that there was some recent patches touching the buspirate.
it can be because of cable length: are the buspirate probe cables too
long for this? [9] I'm not sure...but with the following setup: http://www.flashrom.org/images/6/65/Serduino_laptop.jpeg I had some trouble with reflashing the BIOS and flashrom retried once.
However it can also be the wiring, for instance is it correctly wired? is the write protection on? (it's active-low so it should be high to prevent write-protection...), is the chip select connected to the buspirate?
you could also check the datasheet of your chip...
Denis.
Finally, I managed to have it working.
To summarize what I did: - I checked it all to match your advices - updated bus pirate firmware to v6.1 (I couldn't find the 6.2 update for the bus pirate 3.5 hardware) - replaced some connectors of the probe cable so they stay in place - got the latest flashrom svn sources
Le lundi 18 juin 2012 à 14:35 +0200, Denis 'GNUtoo' Carikli a écrit :
Also Note that there was some recent patches touching the buspirate.
- then I decided to apply these patches on top of the fresh flashrom sources - though, it still didn't work - for some reason, I plugged the motherboard power supply connectors in, just to give it a try (whereas I got them all off the board to make sure it didn't interfere before) - then it worked (see attached log).
Note that with the version of flashrom that with in the OS repos (it was v0.9.2-r1028 from Trisquel GNU/Linux 4.0), it didn't work. Though that is probably because it's outdated: a fresh v0.9.5.2-r1546, even without the recent bus pirate patches applies, worked like a charm!
At 2Mhz, reading the whole 8MB chip took ~20 minutes.
I think next step is to try writing another firmware to the chip via SPI, to see how it goes.
-- Paul Kocialkowski
On Wed, 2012-06-20 at 18:11 +0200, Paul Kocialkowski wrote:
I think next step is to try writing another firmware to the chip via SPI, to see how it goes.
I extracted one from the latest windows binary MSI provides to update the BIOS and will do the flashing later today.
While reading the board user manual, I saw that MSI provides the pinout for the JCOM1 on-board connector: I assumed that's a serial port, which could help when attempting to port coreboot to the board.
Here is the pinout of both the JCOM1 connector and the DB9 connector I used: http://download.paulk.fr/coreboot/msi-z68ma-g45/jcom1-db9-pinout.png
After soldering it all [1], I put the connector on the board [2] and it worked! I tested with a RS232 to USB adapter on one side and getty listening on ttyS0 on the MSI Z68MA-G45 side: I was able to use the shell from the USB adapter side!
I'm not particularly asking for help there, I just thought it would be a good idea to publish all this, in case one wants to start a port to a similar MSI board.
Though, there is something that could be serial-related: I can't find any reference to the SuperIO! coreboot's superiotool didn't detect any. Is it possible that the Z68 chipset is handling serial too?
[1] http://download.paulk.fr/coreboot/msi-z68ma-g45/jcom1-soldered-connector.jpg [2] http://download.paulk.fr/coreboot/msi-z68ma-g45/jcom1-on-board-connector.jpg
-- Paul Kocialkowski
Here is a log of in-system flashing (it worked "out of the box"), for the MSI Z68MA-G45 board! I think it can be added to the list of supported boards.
I'll test SPI flashing with bus pirate or arduino tomorrow.
-- Paul Kocialkowski
On Fri, 22 Jun 2012 21:35:38 +0000 Paul Kocialkowski paulk@paulk.fr wrote:
Here is a log of in-system flashing (it worked "out of the box"), for the MSI Z68MA-G45 board! I think it can be added to the list of supported boards.
I'll test SPI flashing with bus pirate or arduino tomorrow.
Hello Paul,
thanks for your report! I have added your board to our list of supported hardware and will commit that later together with other small changes.
-
Denis 'GNUtoo' Carikli -
Paul Kocialkowski -
Stefan Tauner