Hi,

thanks for your report. I'll focus on the compilation issues you found.
Stefan Tauner already answered the erase related question.

Am 09.09.2011 22:10 schrieb Rohit Vijapure:

> I am trying to mass update the BIOS on MB899 Intel ® CoreTM 2 Duo/CoreTM
> Duo/ Solo 945GM Mini-ITX Motherboard. I did compile the flashrom by
> following two modifications as I was getting errors for the global
> declaration .
>
> *Compilation errors*
> programmer.h:567: warning: declaration of 'programmer' shadows a global
> declaration

gcc bug. It complains about the parameter name in a function declaration
in programmer.h:
void register_spi_programmer(const struct spi_programmer *programmer);
Will add a workaround (rename programmer to pgm here).

> board_enable.c: In function 'nvidia_mcp_gpio_set':
> board_enable.c:892: error: too few arguments to function 'pci_get_dev'
>

> *2. board_enable.c*

>     dev = pci_get_dev(pacc, dev->domain, dev->bus, dev->dev, 1);
> /*    dev = pci_get_dev(pacc, dev->bus, dev->dev, 1);*/

We explicitly check if we need to use the variant with or without the
domain parameter. This check for PCI_LIB_VERSION >= 0x020200 should work
fine.

> [root@ME2000 /tmp]$ flashrom -w he76.052008
> flashrom v0.9.4-r1395 on Linux 2.6.18.6-2g2g (i686), built with libpci
> 2.1.99-test8, GCC 4.0.2 20051125 (Red Hat 4.0.2-8), little endian

And here's the source of the compilation problem: Your distribution
ships a beta version of libpci/pciutils which uses the interface of
libpci 2.2 but without the matching version definition. We could hack
around that with an additional check #if defined(PCI_CAP_ID_HT), but
that matches only 2.1.99-test6 and above which would fix your case but
leave 2.1.99-test3 to 2.1.99-test5 unfixed. It's the best possible
solution, though.

Apparently Red Hat had forgotten the gcc 2.96 disaster and tried to
repeat it with a beta libpci. Oh well.

Regards,
Carl-Daniel

--
http://www.hailfinger.org/