On Fri, Mar 31, 2017 at 09:03:15AM +0200, Sascha Hauer wrote: > When the circular backend searches for the last page written in the > eraseblock, it iterates backwards pagewise from the end of the block. > This is ok for NAND flash, but on NOR flash, which does not have pages, > the code ends up iterating bytewise backwards, calling into mtd each > time. This is very time consuming, so optimize this by reading the whole > eraseblock once and just iterate over the buffer in memory. > > Signed-off-by: Sascha Hauer <s.hauer@xxxxxxxxxxxxxx> > --- > common/state/backend_bucket_circular.c | 18 ++++++++---------- > 1 file changed, 8 insertions(+), 10 deletions(-) > > diff --git a/common/state/backend_bucket_circular.c b/common/state/backend_bucket_circular.c > index 0bce900d0d..53c2aae803 100644 > --- a/common/state/backend_bucket_circular.c > +++ b/common/state/backend_bucket_circular.c > @@ -379,26 +379,24 @@ static int state_backend_bucket_circular_init( > int sub_offset; > uint32_t written_length = 0; > uint8_t *buf; > + int ret; > > - buf = xmalloc(circ->writesize); > + buf = xmalloc(circ->max_size); > if (!buf) > return -ENOMEM; > > + ret = state_mtd_peb_read(circ, buf, 0, circ->max_size); > + if (ret && ret != -EUCLEAN) > + return ret; We leak buf here. But ownership of buf is vague. In one implementation of state_mtd_peb_read() we free buf in case of errors, but not in all paths. This is the version of state_mtd_peb_read() outside the #ifdef __BAREBOX__ block. The version of state_mtd_peb_read() inside the #ifdef __BAREBOX__ will not free buf, so there ownership is more clear. But in both cases we may leak buf. This is as such unrelated to this patch - but noticed while browsing these patches. Sam _______________________________________________ barebox mailing list barebox@xxxxxxxxxxxxxxxxxxx http://lists.infradead.org/mailman/listinfo/barebox