I'm still not really fond of this approach. Repurposing CONFIG_JFFS2 to add more and more filesystems is not a good approach. I don't recommend merging this as is.
Yes it is not good approach. The whole filesystem handling needs to redesign. I think that we can start with discuss about this.
Ideally, the whole filesystem handling should be reworked into a single API, but I understand that it's a non-trivial amount of work. However, at the very least, ROMFS should be configured with a separate CONFIG_xxx macro.
OK. I think that for now I create macro for CONFIG_ROMFS and I add it to cmd_jffs.c file
More comments below.
diff --git a/common/cmd_jffs2.c b/common/cmd_jffs2.c index 513a226..3e6061a 100644 --- a/common/cmd_jffs2.c +++ b/common/cmd_jffs2.c @@ -85,7 +85,7 @@ */
/*
- JFFS2/CRAMFS support
*/
- JFFS2/CRAMFS/ROMFS support
#include <common.h> #include <command.h> @@ -175,6 +175,11 @@ extern int cramfs_load (char *loadoffset, struct
part_info *info, char *filename
extern int cramfs_ls (struct part_info *info, char *filename); extern int cramfs_info (struct part_info *info);
+extern int romfs_check (struct part_info *info); +extern int romfs_load (char *loadoffset, struct part_info *info, char
*filename);
+extern int romfs_ls (struct part_info *info, char *filename); +extern int romfs_info (struct part_info *info);
Bad form and dangerous. Use a header file. (Yes I know this file already uses a bad pattern, but don't follow the example)
OK
static struct part_info* jffs2_part_info(struct mtd_device *dev, unsigned int
part_num);
/* command line only routines */ @@ -1874,14 +1879,22 @@ int do_jffs2_fsload(cmd_tbl_t *cmdtp, int flag, int
argc, char *argv[])
if ((part = jffs2_part_info(current_dev, current_partnum))){
/* check partition type for cramfs */fsname = (cramfs_check(part) ? "CRAMFS" : "JFFS2");
/* check partition type for JFFS2, cramfs, romfs */if (cramfs_check(part)) {fsname = "CRAMFS";} else if (romfs_check(part)) {fsname = "ROMFS";} else {fsname = "JFFS2";}
This will get unwieldy in a real hurry. This should probably be reworked into a table of filesystem handlers which binds all the driver callbacks.
I will add CONFIG_ROMFS macro and it is enough for now.
M