
Hi Simon,
On Wed, 11 Dec 2024 at 15:54, Simon Glass sjg@chromium.org wrote:
At present efi_memory uses struct efi_mem_desc for each node of its memory list. This is convenient but is not ideal, since:
- it means that the fields are under a 'desc' sub-structure
- it includes an unused 'reserved' field
That's interesting. The EFI spec does not define that and I looked as back as 2.5. Heinrich, this predates my involvement with EFI, do you remember why it was added?
Because it's defined in
- it includes virtual_start which is confusing as it is always the same as physical_start
- we must use u64 to store pointers, since that is how they are returned by calls to efi_get_memory_map()
I don't understand the 'pointers' again here. It's a physical address.
As a first step to tidying this up, create a new, private struct to hold these fields.
Signed-off-by: Simon Glass sjg@chromium.org
(no changes since v2)
Changes in v2:
- Add new patch to use a separate stuct for memory nodes
lib/efi_loader/efi_memory.c | 49 +++++++++++++++++++++++++++++++------ 1 file changed, 42 insertions(+), 7 deletions(-)
diff --git a/lib/efi_loader/efi_memory.c b/lib/efi_loader/efi_memory.c index 4bc6d12a1fe..db40aee8353 100644 --- a/lib/efi_loader/efi_memory.c +++ b/lib/efi_loader/efi_memory.c @@ -27,9 +27,39 @@ DECLARE_GLOBAL_DATA_PTR;
efi_uintn_t efi_memory_map_key;
+/**
- struct priv_mem_desc - defines an memory region
- Note that this struct is for use inside U-Boot and is not visible to the
- EFI application, other than through calls to efi_get_memory_map(), where this
- internal format is converted to the external struct efi_mem_desc format.
- @type (enum efi_memory_type): EFI memory-type
- @reserved: unused
- @physical_start: Start address of region in physical memory
- @virtual_start: Start address of region in physical memory
- @num_pages: Number of EFI pages this record covers (each is EFI_PAGE_SIZE
bytes)
- @attribute: Memory attributes (see EFI_MEMORY...)
- */
+struct priv_mem_desc {
u32 type;
u32 reserved;
efi_physical_addr_t physical_start;
efi_virtual_addr_t virtual_start;
u64 num_pages;
u64 attribute;
+};
This is not private, it's described by the EFI spec.
+/**
- struct efi_mem_list - defines an EFI memory record
- @link: Link to prev/next node in list
- @desc: Memory information about this node
- */
struct efi_mem_list { struct list_head link;
struct efi_mem_desc desc;
struct priv_mem_desc desc;
};
#define EFI_CARVE_NO_OVERLAP -1 @@ -116,7 +146,7 @@ static int efi_mem_cmp(void *priv, struct list_head *a, struct list_head *b)
- @desc: memory descriptor
- Return: end address + 1
*/ -static uint64_t desc_get_end(struct efi_mem_desc *desc) +static uint64_t desc_get_end(struct priv_mem_desc *desc) { return desc->physical_start + (desc->num_pages << EFI_PAGE_SHIFT); } @@ -138,8 +168,8 @@ static void efi_mem_sort(void) while (merge_again) { merge_again = false; list_for_each_entry(lmem, &efi_mem, link) {
struct efi_mem_desc *prev;
struct efi_mem_desc *cur;
struct priv_mem_desc *prev;
struct priv_mem_desc *cur; uint64_t pages; if (!prevmem) {
@@ -194,11 +224,11 @@ static void efi_mem_sort(void)
- to re-add the already carved out pages to the mapping.
*/ static s64 efi_mem_carve_out(struct efi_mem_list *map,
struct efi_mem_desc *carve_desc,
struct priv_mem_desc *carve_desc, bool overlap_conventional)
{ struct efi_mem_list *newmap;
struct efi_mem_desc *map_desc = &map->desc;
struct priv_mem_desc *map_desc = &map->desc; uint64_t map_start = map_desc->physical_start; uint64_t map_end = map_start + (map_desc->num_pages << EFI_PAGE_SHIFT); uint64_t carve_start = carve_desc->physical_start;
@@ -678,7 +708,12 @@ efi_status_t efi_get_memory_map(efi_uintn_t *memory_map_size, /* Return the list in ascending order */ memory_map = &memory_map[map_entries - 1]; list_for_each_entry(lmem, &efi_mem, link) {
*memory_map = lmem->desc;
What's wrong with that ?
memory_map->type = lmem->desc.type;
memory_map->reserved = lmem->desc.reserved;
memory_map->physical_start = lmem->desc.physical_start;
memory_map->virtual_start = lmem->desc.virtual_start;
memory_map->num_pages = lmem->desc.num_pages;
memory_map->attribute = lmem->desc.attribute; memory_map--; }
-- 2.34.1
The only problem I see here is the reserved field. I think this patch should just remove that
Thanks /Ilias