Hi, the other day I got a 3C905CX-TX-M NIC.
I don't have the computer I tested it on up and running right now, but it is non-functional. Flashrom can not detect the AT49BV512 on it. A forced read delivers all 0xFF's, which I assume it is pulling out of the ether. I guess it *could* be blank; It seems unlikely to not autodetect and function on a forced read though.
Anyways, this wouldn't be an issue, but the PCI device ID is the same as the 3C905C-TX [dev_id 0x9200] (*iirc - it might be one of the other supported 3C905 NICs*), so flashrom reports it as a supported NIC.
I guess there should be a note somewhere, specifying that flashrom doesn't support 3com NICs with the newer, smaller, BGA package ASIC.
Here's a photo of the card. On mine the flash is soldered on, but I believe I've seen socketed versions in the past. If they're all soldered on, I guess it doesn't matter much - apart from saving someone the time of soldering on a socket.
http://electrontube.org/open/3com_mod/3C905CX-TX-M.jpg
I'll try and get that computer back up this weekend and post the output.
Mark
Hi,
On Fri, Sep 18, 2009 at 06:38:27AM -0600, Marko Kraljevic wrote:
Hi, the other day I got a 3C905CX-TX-M NIC.
I don't have the computer I tested it on up and running right now, but it is non-functional. Flashrom can not detect the AT49BV512 on it. A forced read delivers all 0xFF's, which I assume it is pulling out of the ether. I guess it *could* be blank; It seems unlikely to not autodetect and function on a forced read though.
Hm, please send the full "flashrom -p nic3om -V" output to the list.
Also, what is the "lspci -xxxvvv" output?
Anyways, this wouldn't be an issue, but the PCI device ID is the same as the 3C905C-TX [dev_id 0x9200] (*iirc - it might be one of the other supported 3C905 NICs*), so flashrom reports it as a supported NIC.
I guess there should be a note somewhere, specifying that flashrom doesn't support 3com NICs with the newer, smaller, BGA package ASIC.
That's unrelated, the BGA package doesn't affect the flashing capabilities in any way.
However, the chip may indeed be empty. I just used flashrom r731 on a 3Com NIC with BGA packaged chip, which also had an empty chip:
$ ./flashrom -p nic3com -r nic.dd flashrom v0.9.1-r731 Found "3COM 3C905C: EtherLink 10/100 PCI (TX)" (10b7:9200, BDF 05:04.0). Calibrating delay loop... OK. Found chip "Atmel AT49BV512" (64 KB, Parallel) at physical address 0xffff0000. Reading flash... done.
$ hexdump nic.dd 0000000 ffff ffff ffff ffff ffff ffff ffff ffff * 0010000
However, writing a random image into the chip works fine, so the chip was simply empty in my case indeed.
Here's a photo of the card. On mine the flash is soldered on, but I believe I've seen socketed versions in the past. If they're all soldered on, I guess it doesn't matter much - apart from saving someone the time of soldering on a socket.
Yep, soldered or socketed doesn't matter here. As it's a NIC any misflashed images can easily be fixed, the card is not automatically and terminally "bricked" (like it's the case with mainboards with soldered chips).
I'll try and get that computer back up this weekend and post the output.
Yep, please do.
Thanks, Uwe.
Well scratch that.
I put my flashing machine back together, and installed two 3C905CX-TX-M cards, and they are now autodetecting their flash chips; I have no idea why it didn't work the other day.
I think I'll try installing a socket tomorrow and see if it will work with larger chips ( i'm unsure whether the >512kbit addr line(s) are drawn out to the flash or not ).
On Wed, Sep 23, 2009 at 08:39:01PM -0600, Marko Kraljevic wrote:
Well scratch that.
I put my flashing machine back together, and installed two 3C905CX-TX-M cards, and they are now autodetecting their flash chips; I have no idea why it didn't work the other day.
I think I'll try installing a socket tomorrow and see if it will work with larger chips ( i'm unsure whether the >512kbit addr line(s) are drawn out to the flash or not ).
The maximum supported chip size as per 3Com datasheets is 128KB.
Uwe.
On 25.09.2009 01:55, Uwe Hermann wrote:
On Wed, Sep 23, 2009 at 08:39:01PM -0600, Marko Kraljevic wrote:
I put my flashing machine back together, and installed two 3C905CX-TX-M cards[...] ( i'm unsure whether the >512kbit addr line(s) are drawn out to the flash or not ).
The maximum supported chip size as per 3Com datasheets is 128KB.
Are there any GPIOs one could wire up as additional address lines with special handling code? With the current flashrom infrastructure, that would be almost trivial.
Regards, Carl-Daniel
Carl-Daniel Hailfinger wrote:
On 25.09.2009 01:55, Uwe Hermann wrote:
On Wed, Sep 23, 2009 at 08:39:01PM -0600, Marko Kraljevic wrote:
I put my flashing machine back together, and installed two 3C905CX-TX-M cards[...] ( i'm unsure whether the >512kbit addr line(s) are drawn out to the flash or not ).
The maximum supported chip size as per 3Com datasheets is 128KB.
Are there any GPIOs one could wire up as additional address lines with special handling code? With the current flashrom infrastructure, that would be almost trivial.
Regards, Carl-Daniel
I don't believe there are. A while back, on a different NIC, I put a latch on the upper address lines; I'm quite sure I made a post about it, but maybe not. It was pretty ugly, but worked - planning to get a PCB made someday; should be able to just drop it in the socket on a unmodified card :-)
The top address line selects a latch, and then you can write the upper address bits to the latch...
schematic: http://electrontube.org/3com_mod/minimal_sm.png
code change... could be beautified some: http://electrontube.org/3com_mod/3comhack.diff
picture of the hardware: http://electrontube.org/3com_mod/etherlink_mod.jpg
I was planning on using the leftover GPIO's off the latch to enable HV for 27xxx EPROMs. Someday...
-mark
-
Carl-Daniel Hailfinger -
Marko Kraljevic -
Uwe Hermann