fix for mtd partitions for erase_size != 2^X

Here a trivial patch,
which allows correct creation of mtd partitions, which have erase sizes which are not a power of 2.

Best regards
Peter

Sorry, I didn't succeed to get the patch accepted by the mailing list as attachment.

mtd/mtdpart.c 2005-09-23 09:29:37.000000000 +0200 renamed from mtd/mtdpart.c.orig 2005-09-23 09:28:07.000000000 +0200 +3 -3

@@ -465,9 +465,9 @@
         if (slave->offset == MTDPART_OFS_APPEND)
             slave->offset = cur_offset;
         if (slave->offset == MTDPART_OFS_NXTBLK) {
-            u_int32_t emask = master->erasesize-1;
-            slave->offset = (cur_offset + emask) & ~emask;
-            if (slave->offset != cur_offset) {
+            slave->offset = cur_offset;
+            if ((cur_offset % master->erasesize) != 0) {
+                slave->offset = ((cur_offset / master->erasesize) + 1) 
 * master->erasesize;
                 printk(KERN_NOTICE "Moving partition %d: "
                        "0x%08x -> 0x%08x\n", i,

                        cur_offset, slave->offset);

Maybe this:
#define TRUE_ALIGN(x,a) (((x)+(a)-1) - ((x)+(a)-1) % a)

The macro is still complicated, but we just have to get it right once.

Jörn

Quoted 9 lines · Peter Menzebach

Here a trivial patch,
which allows correct creation of mtd partitions, which have erase sizes which are not a power of 2.

Best regards
Peter

Sorry, I didn't succeed to get the patch accepted by the mailing list as attachment.

I actually prefer inline. Except that your mailer seems to have messed things up - grr.

Quoted 15 lines

mtd/mtdpart.c 2005-09-23 09:29:37.000000000 +0200 renamed from mtd/mtdpart.c.orig 2005-09-23 09:28:07.000000000 +0200 +3 -3

@@ -465,9 +465,9 @@
        if (slave->offset == MTDPART_OFS_APPEND)
            slave->offset = cur_offset;
        if (slave->offset == MTDPART_OFS_NXTBLK) {
-            u_int32_t emask = master->erasesize-1;
-            slave->offset = (cur_offset + emask) & ~emask;
-            if (slave->offset != cur_offset) {
+            slave->offset = cur_offset;
+            if ((cur_offset % master->erasesize) != 0) {
+                slave->offset = ((cur_offset / master->erasesize) + 1) 
 * master->erasesize;
                printk(KERN_NOTICE "Moving partition %d: "
                       "0x%08x -> 0x%08x\n", i,

Quoted 1 line

                       cur_offset, slave->offset);

Looks complicated. I guess we could use the equivalent of this:
#define ALIGN(x,a) (((x)+(a)-1)&~((a)-1))
in some header. Then your patch would be as simple as

         if (slave->offset == MTDPART_OFS_APPEND)
             slave->offset = cur_offset;
         if (slave->offset == MTDPART_OFS_NXTBLK) {

mtd/mtdpart.c 2005-09-23 09:29:37.000000000 +0200 renamed from mtd/mtdpart.c.orig 2005-09-23 09:28:07.000000000 +0200 +1 -2

-            u_int32_t emask = master->erasesize-1;
-            slave->offset = (cur_offset + emask) & ~emask;
+	     slave->offset = TRUE_ALIGN(cur_offset, master->erasesize);
             if (slave->offset != cur_offset) {
                 printk(KERN_NOTICE "Moving partition %d: "
                        "0x%08x -> 0x%08x\n", i,

Maybe this:
#define TRUE_ALIGN(x,a) (((x)+(a)-1) - ((x)+(a)-1) % a)

The macro is still complicated, but we just have to get it right once.

Jörn

Quoted 1 line

Here the patch again. As the macro does not seem to reduce the complexity of the line seriously and would be used only once, I did put in a comment, which explains inline, what should go on there.