On 17.07.2010 14:44, Michael Karcher wrote:
Am Samstag, den 17.07.2010, 14:35 +0200 schrieb Carl-Daniel Hailfinger:
Signed-off-by: Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net
This patch as-is is Acked-by: Michael Karcher flashrom@mkarcher.dialup.fu-berlin.de
Thanks, committed in r1085.
According to current flashrom desing I see no technical reason to not accept the patch.
BUT: We really should think about killing the SPI bitbang master table. Why not just pass a pointer to a "struct bitbang_spi_master_enty" to bitbang_spi_init which will then be stored in a static global pointer instead of a table index as it is done currently? Maybe it's the best time to do this change now before we merge the bit-bang drivers if this change is accepted.
This design question applies to all programmer related tables. Advantages: The current checks for miscompilation on startup would be obsolete, and the programmer enum and related variables would not be needed anymore either. Disadvantages: It will be a lot more difficult to blacklist certain actions based on the programmer (because that action is not supported) and that may require conversion of some error avoidance paths to error recovery paths.
Not sure. It looked good at a first glance, but now I have serious doubts.
Regards, Carl-Daniel