On Mon, Dec 23, 2013 at 3:26 PM, Hauke Mehrtens <ha...@hauke-m.de> wrote: > On 12/21/2013 07:52 PM, Rafał Miłecki wrote: >> 2013/12/5 Cody P Schafer <de...@codyps.com>: >>> F7D3301/F7D7301, F7D3302/F7D7302, and F7D4301 replace the "HDR0" with a >>> custom byte sequence (which appears to be a date). Also, they all >>> appear to support a single "QA" byte sequence. >>> >>> Just support the "QA" byte sequence, we don't need the others. >> >> Your patch seems to do the opposite by supporting all magics, not just >> "QA". I'm little afraid of that. As you noted by yourself, these >> magics are probably just dates. I'm afraid of having one magic per >> device or even more. >> >> So why your patch actually adds support for detecting all these MAGICs >> instead of just QA (like noted in the commit message)? > > I think firmware upgrade over the Web administration interface only > works with these dates as magic and not with this QA, that works only > over CFE, the commit message should be fixed. > > I am a little frightened that we would detected the wrong bytes as a trx > header and then apply a wrong partition design.
The obvious solution to that would be to check the crc32 of the trx header, which is AFAICT not currently done. But that would mean rewriting the current scan method. I think it's quite unlikely that the crc32 matches without it being a trx header. This is already being done by the mtd utility's trx fixups, btw. Jonas _______________________________________________ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
