On Sun, Sep 22, 2013 at 12:21 AM, Sascha Silbe t-uboot@infra-silbe.de wrote:
Hi Jagannadha,
Jagannadha Sutradharudu Teki jagannadha.sutradharudu-teki@xilinx.com writes:
Added new spi_flash_probe support, currently added N25Q* flash part attributes support.
[...]
I've given your series another try on Cubox. It looks better this time (e.g. the probe issue is gone), but there are still a couple of issues. I'm going to reply to each of the affected patches (2, 7, 14) individually so it shows up properly in Patchwork.
This patch causes a major regression and thus breaks bisectability. It rips out support for all SPI flash chips except for STMICRO ones, unless CONFIG_SPI_FLASH_LEGACY is defined. However, CONFIG_SPI_FLASH_LEGACY doesn't get defined anywhere.
Ok.
I think for better bisect-ability, I will keep the old probe as spi_flash_probe_legacy.c which is defined in CONFIG_SPI_FLASH and new anyway spi_flash-probe.c which will defined in CONFIG_SPI_FLASH once all flash probes are merged.
I think it could be a better idea, let me know if you have any comments.
A better approach would be to make the new code opt-in rather than opt-out, and selecting it automatically for CONFIG_SPI_FLASH_<vendor> as support for each vendor is added.
PS: What's the etiquette for the CC list of replies to patches? Keep all original recipients CC'ed? Or should I have dropped the chip vendor contacts for this reply?
Yes, you could add CCed one on reply, i think it's better to know the flash vendors to know how u-boot is supporting sf. My intention of doing this, if any flash vendor is interesting about these features in u-boot and they might do a test on their hw.