Hi Ilias,
On Wed, 4 Dec 2024 at 07:46, Ilias Apalodimas ilias.apalodimas@linaro.org wrote:
Hi Simon,
On Sun, Dec 01, 2024 at 08:24:19AM -0700, Simon Glass wrote:
The EFI-loader implementation converts things back and forth between addresses and pointers, with not much consistency in how this is done.
Within most of U-Boot a pointer is a void * and an address is a ulong
This convention is very helpful, since it is obvious in common code as to whether you need to call map_sysmem() and friends, or not.
As part of cleaning up the EFI memory-management, I found it almost impossible to know in some cases whether something is an address or a pointer. I decided to give up on that and come back to it when this is resolved.
This series starts applying the normal ulong/void * convention to the EFI_loader code, so making things easier to follow. For now, u64 is often used instead of ulong, but the effect is the same.
The main changes are:
- Rather than using the external struct efi_mem_desc data-structure for internal bookkeeping, create a new one, so it can have different semantics
- Within efi_memory.c addresses are used, rather than addresses masquerading as pointers. The conversions are done in efi_boottime
Link: https://lore.kernel.org/u-boot/20240725135629.3505072-1-sjg@chromium.org/
In v3 I did ask you to run SCT and spent some time on IRC to explain how to run it. The existing code passes all tests related to memory allocations. This patch introduced 99 failures in MemoryAllocationServicesTest. Apart from all the other feedback, SCT must pass.
Hmm I must be doing something wrong, as it passed for me. I'll try again.
Regards, Simon