Hello everybody, especially board maintainers!
I just comitted a bigger patch that implements a new "mtdparts" command and a common partition scheme that allows U-Boot and Linux to share the same idea about flash partitioning.. Old, obsolete and duplicated code was cleaned up and replaced by the new partitioning method. There are two possible approaches now:
The first one is to define a single, static partition:
#undef CONFIG_JFFS2_CMDLINE #define CONFIG_JFFS2_DEV "nor0" #define CONFIG_JFFS2_PART_SIZE 0xFFFFFFFF /* use whole device */ #define CONFIG_JFFS2_PART_SIZE 0x00100000 /* use 1MB */ #define CONFIG_JFFS2_PART_OFFSET 0x00000000
The second method is to use the "mtdparts" command line and dynamic partitioning:
/* mtdparts command line support */ #define CONFIG_JFFS2_CMDLINE #define MTDIDS_DEFAULT "nor1=zuma-1,nor2=zuma-2" #define MTDPARTS_DEFAULT "mtdparts=zuma-1:-(jffs2),zuma-2:-(user)"
Command line of course produces bigger images, and may be inappropriate for some targets, so by default it's off.
The code was tested on the following targets: Alaska8220, CPCI4052, NC650, PPChamelion, incaip.
Besides the above targets, there were several other targets affected, where necessary changes were made. However, no testing was done for those targets, only compilation was verified (for PPC, MIPS and ARM targets). Due to the lack of tool chain for NICOS and i386 I was not able to compile the following targets: ADNPESC1, DK1C20, DK1S10, sc520_cdp, sc520_spunk.
Among the modified and not tested targets there is a group of targets which were implementing their custom multi-partition scheme. This was mostly duplicating jffs2_part_info() routine code. Introducing the new common scheme required removing this duplicated code, so those targets should be especially carefully tested by the board main- tainers. Target list that require special attention: xsengine, voiceblue, mx1fs2, MPC8260ADS, innokom and VoVPN-GW (JFFS2 is by default not compiled in) and especially SXNI855T where the jffs2_part_info() routine was doing some custom NAND read caching.
Thanks.
Best regards,
Wolfgang Denk