On 11/22/2012 05:16 AM, Lukasz Majewski wrote:
Hi Piotr,
Dear Stephen,
Stephen Warren wrote at Monday, November 19, 2012 9:17 PM: On 11/09/2012 03:48 AM, Piotr Wilczek wrote:
The restoration of GPT table (both primary and secondary) is now possible. Simple GUID generation is supported.
+void gpt_fill_header(block_dev_desc_t *dev_desc, gpt_header *gpt_h)
- gpt_h->first_usable_lba = cpu_to_le64(34);
- gpt_h->last_usable_lba = cpu_to_le64(dev_desc->lba - 34);
Is lba the number of LBAs or the last LBA number? include/part.h says its the number of LBAs, in which case I think that should be dev_desc-
lba - 1 - 34?
LBA is the number of blocks. So the addressable range is 0 to LBA-1. The last_usable_lba is calculated with following equation (it is similar with parted tool):
lba - 2 - ptes_size = lba -2 -32 = lba - 34
ptes_size = number_of_entries(128) * sizeof(pte_entry) (128) / sector_size (512) = 32
Ah right, I was forgetting that there is no protective MBR at the end of the disk, so 1 less sector is used there.
- s = getenv("uuid_gpt_disk");
I'd rather not depend on environment variables; can't this be a command-line parameter to the gpt command?
I think, that the uuid_gpt_disk shall be added as another key=value to the partitions.
e.g. gpt uuid_gpt_disk=6aa01a70-349d-11e2-ae74-001fd09285c0;name=boot,size=100MiB,uuid=xx;name=kernel, size=200MiB,uuid=yy;....
I guess that'd work. Note that means the calculation of # of partitions based on counting the # of ; different. Perhaps a separate command-line argument would be simpler?