On 16 December 2014 at 16:30, Imre Kaloz <ka...@openwrt.org> wrote: > On Tue, 16 Dec 2014 16:06:27 +0100, Gergely Kiss <mail.g...@gmail.com> > wrote: >> On 16 December 2014 at 14:39, Flávio Silveira <f...@terra.com.br> wrote: >>> >>> On 16/12/2014 11:23, Gergely Kiss wrote: >>>> I'm not sure I understand what you mean here. Are there any devices >>>> out there with multiple flash chips on-board? As far as I know, UBI >>>> cannot work on top of block devices (eg. USB flashes, SD cards), only >>>> raw flash chips so I can't really imagine what exact use-case you >>>> mean. Could you please clarify? >>>> >>> I think this is where this discussion started and Rafal explains a bit >>> more about the hardware: >>> >>> https://lists.openwrt.org/pipermail/openwrt-devel/2014-December/029717.html >>> > > Rafal, what about mtdconcat?
That would be nice because I could use space available on the serial flash. So far I assumed to ignore that extra ~10 MiB and focus on NAND mostly. However I'm not sure if this will be possible without heavy changes. I have rootfs_data (JFFS2) partition on serial flash. I would need to concat it with my UBI partition on NAND flash. 1) I'm afraid it may be complex to change the way OpenWrt treats rootfs_data. It's coded in many places. 2) Concanation has to be handled in the kernel. There isn't any DT driver I could use for bcm53xx. Guess I'd need to write once 3) I have no idea if UBI can work well on concan-ated partition. In theory yes, but in practice? UBI cares pretty much about wearing I believe. -- Rafał _______________________________________________ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
