Hi All!
Making experiments with Alpha in such a configuration:
DEC AlphaPC 164LX
CPU: 600 MHz 21164 EV56, 2MB BCache
MEM: 1024 MB SDRAM PC133 ECC REG 72bit
HDD: Seagate ST3366007LW 37GB x2
Video: S3 Trio 64V+
NET: Intel 82558 PCI-X 10/100
SCSI: Adaptec 39160, Qlogic 1020
The motherboard is Samsung, apparently it is "Order number 21A4-C1 Montherboard DIGITAL UNIX" (versus 21A4-C0 which is for NT).
SRM is flashed, works ok, Tru64 UNIX is on board and works fine.
But me and my friend want Windows NT, just for fun)
Here we go:
1) getting Arc ROM from here: https://people.freebsd.org/~msmith/alpha/pc164lx/
2) preparing floppy with it
3) booting:
https://pic.maxiol.com/?v=1616249280.2990710549.dsc0162.jpg&dp=2
https://pic.maxiol.com/?v=1616249453.2990710549.dsc0163.jpg&dp=2
https://pic.maxiol.com/?v=1616249543.2990710549.dsc0164.jpg&dp=2
https://pic.maxiol.com/?v=1616249698.2990710549.dsc0165.jpg&dp=2
Further investigations lead us to the flash chip difference:
E28F008S5 is on board, while Arc BIOS (aka AlphaBIOS) knows only 28F008SA.
Chip ids are 89A6 and 89A2 respectively.
Well, next we dumbly tried to patch the AlphaBIOS image, replacing 89 A2 with 89 A6.
Checksum mismatch, as was expected;
checksum patched (offset 0C);
reboot, checksum is ok, but NVRAM error remains the same.
Meanwhile I opened ROM image in IDA Pro with Alpha support. Quick investigation shows this:
- both SRM and Arc have mostly identical loader in the first ~5% of the image
- reduced prime numbers table next to it (with gaps, looks like logarithmic scale)
- further code could not be disassembled. First I thought the ROM is packed/encrypted, now I think it is what is called PALCode - another DEC microcode (IDA does not recognize it)
- almost no strings, except version control ids (not much helpful)
No clue further, except re-soldering the flash chip, which we wouldn't want to do without a reason.
Well, the questions are:
1) any PALCode description and hints about it translating it back to human friendly form
2) ready solution? lol )
Making experiments with Alpha in such a configuration:
DEC AlphaPC 164LX
CPU: 600 MHz 21164 EV56, 2MB BCache
MEM: 1024 MB SDRAM PC133 ECC REG 72bit
HDD: Seagate ST3366007LW 37GB x2
Video: S3 Trio 64V+
NET: Intel 82558 PCI-X 10/100
SCSI: Adaptec 39160, Qlogic 1020
The motherboard is Samsung, apparently it is "Order number 21A4-C1 Montherboard DIGITAL UNIX" (versus 21A4-C0 which is for NT).
SRM is flashed, works ok, Tru64 UNIX is on board and works fine.
But me and my friend want Windows NT, just for fun)
Here we go:
1) getting Arc ROM from here: https://people.freebsd.org/~msmith/alpha/pc164lx/
2) preparing floppy with it
3) booting:
https://pic.maxiol.com/?v=1616249280.2990710549.dsc0162.jpg&dp=2
https://pic.maxiol.com/?v=1616249453.2990710549.dsc0163.jpg&dp=2
https://pic.maxiol.com/?v=1616249543.2990710549.dsc0164.jpg&dp=2
https://pic.maxiol.com/?v=1616249698.2990710549.dsc0165.jpg&dp=2
Further investigations lead us to the flash chip difference:
E28F008S5 is on board, while Arc BIOS (aka AlphaBIOS) knows only 28F008SA.
Chip ids are 89A6 and 89A2 respectively.
Well, next we dumbly tried to patch the AlphaBIOS image, replacing 89 A2 with 89 A6.
Checksum mismatch, as was expected;
checksum patched (offset 0C);
reboot, checksum is ok, but NVRAM error remains the same.
Meanwhile I opened ROM image in IDA Pro with Alpha support. Quick investigation shows this:
- both SRM and Arc have mostly identical loader in the first ~5% of the image
- reduced prime numbers table next to it (with gaps, looks like logarithmic scale)
- further code could not be disassembled. First I thought the ROM is packed/encrypted, now I think it is what is called PALCode - another DEC microcode (IDA does not recognize it)
- almost no strings, except version control ids (not much helpful)
No clue further, except re-soldering the flash chip, which we wouldn't want to do without a reason.
Well, the questions are:
1) any PALCode description and hints about it translating it back to human friendly form
2) ready solution? lol )