01:29 <crb> god in heaven, trying to figure out make files makes me want to poke my eyes with a fork

01:30 <hanetzer> heh

04:11 <xypron> crb: you should read the align value at run time in your EFI code. Some non-U-Boot EFI implementations have been reported to be setting too small a value on the GRUB mailing list. The last patch version for GRUB therefore uses a 512 minimum.

04:16 <crb> xypron sorry I'm a bit confused u-boot seems to be enforcing an alignment requirement and I think the code u-boot is loading (freebsd efi boot loader) is, I think aligning to 512 bytes but this fails which leads me to believe that u-boot is imposing more than 512 byte alignment. So I'm trying to find out what u-boot is requiring. Is there and efi way to query u-boot for the alignment it requires?

04:19 <xypron> crb: the align field is exposed in the block io protocol. U-Boot sets this value to the sector size.

04:20 <crb> xypron, that's what I thought it should be but since it's loading from a microSD card I think that should be 512 bytes

04:23 <xypron> crb: OpenBSD uses AllocatePages. So their buffer is always 4096 aligned.

04:24 <crb> xypron, looking at the freebsd code I see: static char ufs_buffer[BSD_LABEL_BUFFER] __aligned(512);

04:24 <crb> xypron, so I think they're allocating an 8k buffer that's 512 aligned. but the boot fails unless I comment out the alignment check in u-boot

04:26 <xypron> crb: some EDK2 platforms require higher alignment then 512. You should do the alignment according to the align field at runtimr and not compile in a fixed value.

04:27 <xypron> Which platform fails?

04:27 <crb> xypron, I would agree, but they're statically allocating the buffer and forcing alignment. I'll have to look into how I can allocate a buffer instead of statically allocating it. I'm not sure there is a malloc in this limited environment

04:28 <crb> xypron, Xilinx Zynq 700 FreeBSD 13.0

04:28 <xypron> crb: use the EFI API for allocation.

04:29 <crb> xypron, ok, I'll have to look into querring and allocating via EFI, it's all kind new to see

04:29 <crb> xypron, thank you!

04:33 <xypron> crb: In the UEFI specification look for AllocatePages, AllicatePool, FreePages, FreePool. LOADER_DATA should be the right memory type.

04:34 <crb> xypron, I'm looking at AllocatePages now,

04:34 <crb> xypron, also searching for how to query the alignment requirement from the block IO protocol

04:39 <xypron> crb: it is just a structure field of thr protocol.

04:39 <crb> xypron, yep I think I'm seeing that, it just takes make a bit to troll through the standard, it's all very new to me

04:39 <crb> s/make/me

04:55 <crb> and it looks like my ssd has died .... .sad face

05:06 <hanetzer> sadge.