Hi,
On Sat, 1 Jul 2017 15:42:57 -0700 alison@peloton-tech.com wrote:
From: Alison Chaiken alison@peloton-tech.com
Make the partition table available for modification by reading it from the user-specified device into a linked list. Provide an accessor function for command-line testing.
Signed-off-by: Alison Chaiken alison@peloton-tech.com
Changes since v7: The failure-handling logic of get_gpt_info() now is triggered properly when valid_part=1.
cmd/gpt.c | 119 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++ include/part.h | 7 ++++ 2 files changed, 126 insertions(+)
diff --git a/cmd/gpt.c b/cmd/gpt.c index 65fb80b..dd5f78b 100644 --- a/cmd/gpt.c +++ b/cmd/gpt.c
[...]
+static int get_gpt_info(struct blk_desc *dev_desc) +{
- /* start partition numbering at 1, as U-Boot does */
- int valid_parts = 1, p, ret;
- disk_partition_t info;
- struct disk_part *new_disk_part;
- if (disk_partitions.next == NULL)
INIT_LIST_HEAD(&disk_partitions);- for (p = 1; p <= MAX_SEARCH_PARTITIONS; p++) {
ret = part_get_info(dev_desc, p, &info);if (ret)continue;new_disk_part = allocate_disk_part(&info, valid_parts);if (IS_ERR(new_disk_part))goto out;list_add_tail(&new_disk_part->list, &disk_partitions);valid_parts++;- }
- if (valid_parts == 1) {
printf("** No valid partitions found **\n");goto out;- }
- return --valid_parts;
I personally find it somewhat strange to initialize a counter to 1 and then return its value reduced by one, rather than initializing it to 0 and returning the exact counter value.
Lothar Waßmann