On Tue, Nov 11 2025, Anshul Dalal <[email protected]> wrote: >> int part_get_type_by_name(const char *name) >> { >> struct part_driver *drv = >> @@ -322,12 +344,9 @@ int part_get_info_by_type(struct blk_desc *desc, int >> part, int part_type, >> struct disk_partition *info) >> { >> struct part_driver *drv; >> + int ret = -ENOENT; >> >> if (blk_enabled()) { >> - /* The common case is no UUID support */ >> - disk_partition_clr_uuid(info); >> - disk_partition_clr_type_guid(info); >> - >> if (part_type == PART_TYPE_UNKNOWN) { >> drv = part_driver_lookup_type(desc); >> } else { >> @@ -339,18 +358,16 @@ int part_get_info_by_type(struct blk_desc *desc, int >> part, int part_type, >> desc->part_type); >> return -EPROTONOSUPPORT; >> } >> - if (!drv->get_info) { >> - PRINTF("## Driver %s does not have the get_info() >> method\n", >> - drv->name); >> - return -ENOSYS; >> - } >> - if (drv->get_info(desc, part, info) == 0) { >> + >> + ret = part_driver_get_info(drv, desc, part, info); >> + if (ret && ret != -ENOSYS) { >> + ret = -ENOENT; > > Why are we overwriting the err code from part_driver_get_info here?
That's essentially what the code did originally (it only returned 0 in case ->get_info returned 0, otherwise it fell through to the "return -ENOENT" at the bottom). And it turns out that that behaviour is expected by lots of tests; when I inadvertently changed that logic in v2 to actually propagate whatever error ->get_info returned, lots of tests broke: https://lore.kernel.org/u-boot/20251107201927.GA2243313@bill-the-cat/ And this is the only difference between v2 and v3, I tweaked this so it would preserve that "either success or -ENOENT [or -ENOSYS]" behaviour. Rasmus
