public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Ard Biesheuvel <ardb+git@google.com>
To: linux-kernel@vger.kernel.org
Cc: linux-efi@vger.kernel.org, x86@kernel.org,
	 Ard Biesheuvel <ardb@kernel.org>,
	"Mike Rapoport (Microsoft)" <rppt@kernel.org>,
	 Benjamin Herrenschmidt <benh@kernel.crashing.org>,
	Dave Young <ruirui.yang@linux.dev>,
	 Gregory Price <gourry@gourry.net>
Subject: [PATCH v3 16/17] x86/efi: Avoid EFI_MEMORY_RUNTIME for early EFI boot memory reservations
Date: Thu, 23 Apr 2026 17:20:41 +0200	[thread overview]
Message-ID: <20260423152024.1098465-35-ardb+git@google.com> (raw)
In-Reply-To: <20260423152024.1098465-19-ardb+git@google.com>

From: Ard Biesheuvel <ardb@kernel.org>

Currently, memblock reservations of EFI boot services memory made before
all EFI boot services memory is temporarily reserved are upgraded, by
being marked with the EFI_MEMORY_RUNTIME bit, and this results in the
entire region to remain reserved permanently, regardless of the size of
the original memblock reservation that triggered this.

This is a hack, and may be quite inefficient in cases where the firmware
does a good job of merging memory map entries.

So instead, rely on the MEMBLOCK_RSRV_KERN flag, by marking existing
memblock reservations with this flag before creating the new, temporary
ones with the flag cleared. This unifies the treatment of early vs late
memblock reservations inside EFI boot services memory, and avoids
clobbering the EFI memory map.

Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
---
 arch/x86/platform/efi/quirks.c | 66 ++++++--------------
 1 file changed, 18 insertions(+), 48 deletions(-)

diff --git a/arch/x86/platform/efi/quirks.c b/arch/x86/platform/efi/quirks.c
index b7c8337d8f88..fc6a15c2ace6 100644
--- a/arch/x86/platform/efi/quirks.c
+++ b/arch/x86/platform/efi/quirks.c
@@ -273,7 +273,6 @@ void __init efi_reserve_boot_services(void)
 		u64 start = max(md->phys_addr, SZ_1M);
 		u64 end = md->phys_addr + (md->num_pages << EFI_PAGE_SHIFT);
 		u64 size = end - start;
-		bool already_reserved;
 
 		if (end <= start)
 			continue;
@@ -282,37 +281,11 @@ void __init efi_reserve_boot_services(void)
 		    md->type != EFI_BOOT_SERVICES_DATA)
 			continue;
 
-		already_reserved = memblock_is_region_reserved(start, size);
+		/* upgrade existing reservations to MEMBLOCK_RSRV_KERN */
+		if (memblock_is_region_reserved(start, size))
+			memblock_reserved_mark_kern(start, size);
 
-		/*
-		 * Because the following memblock_reserve() is paired
-		 * with free_reserved_area() for this region in
-		 * efi_free_boot_services(), we must be extremely
-		 * careful not to reserve, and subsequently free, critical
-		 * regions of memory that somebody else has already reserved.
-		 */
-		if (!already_reserved) {
-			memblock_reserve(start, size);
-
-			/*
-			 * If we are the first to reserve the region, no
-			 * one else cares about it. We own it and can
-			 * free it later.
-			 */
-			if (can_free_region(start, size))
-				continue;
-		}
-
-		/*
-		 * We don't own the region. We must not free it.
-		 *
-		 * Setting this bit for a boot services region really
-		 * doesn't make sense as far as the firmware is
-		 * concerned, but it does provide us with a way to tag
-		 * those regions that must not be paired with
-		 * memblock_phys_free().
-		 */
-		md->attribute |= EFI_MEMORY_RUNTIME;
+		memblock_reserve(start, size);
 	}
 }
 
@@ -443,25 +416,22 @@ void __init efi_unmap_boot_services(void)
 		 */
 		efi_unmap_pages(md);
 
-		if (!(md->attribute & EFI_MEMORY_RUNTIME)) {
-			/*
-			 * With CONFIG_DEFERRED_STRUCT_PAGE_INIT parts of the memory
-			 * map are still not initialized and we can't reliably free
-			 * memory here.
-			 * Queue the ranges to free at a later point.
-			 */
-			if (efi_add_range_to_free(start, start + size, &has_reservations)) {
-				pr_err("Failed to reallocate storage for freeable EFI regions\n");
-				clear_bit(EFI_MEMMAP, &efi.flags);
-				return;
-			}
-
-			/* Continue without advancing new_md so this region is omitted */
-			if (!has_reservations)
-				continue;
-
+		/*
+		 * With CONFIG_DEFERRED_STRUCT_PAGE_INIT parts of the memory
+		 * map are still not initialized and we can't reliably free
+		 * memory here.
+		 * Queue the ranges to free at a later point.
+		 */
+		if (efi_add_range_to_free(start, start + size, &has_reservations)) {
+			pr_err("Failed to reallocate storage for freeable EFI regions\n");
+			clear_bit(EFI_MEMMAP, &efi.flags);
+			return;
 		}
 
+		/* Continue without advancing new_md so this region is omitted */
+		if (!has_reservations)
+			continue;
+
 		/* Advance new_md so this region is preserved in the EFI memory map */
 		new_md += efi.memmap.desc_size;
 	}
-- 
2.54.0.rc2.544.gc7ae2d5bb8-goog


  parent reply	other threads:[~2026-04-23 15:21 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-04-23 15:20 [PATCH v3 00/17] efi/x86: Avoid the need to mangle the EFI memory map Ard Biesheuvel
2026-04-23 15:20 ` [PATCH v3 01/17] x86/efi: Omit redundant kernel image overlap check Ard Biesheuvel
2026-04-23 15:20 ` [PATCH v3 02/17] x86/efi: Drop redundant EFI_PARAVIRT check Ard Biesheuvel
2026-04-23 15:20 ` [PATCH v3 03/17] x86/efi: Only merge EFI memory map entries on 32-bit systems Ard Biesheuvel
2026-04-23 15:20 ` [PATCH v3 04/17] x86/efi: Defer sub-1M check from unmap to free stage Ard Biesheuvel
2026-04-23 15:20 ` [PATCH v3 05/17] x86/efi: Simplify real mode trampoline allocation quirk Ard Biesheuvel
2026-04-23 15:20 ` [PATCH v3 06/17] x86/efi: Unmap kernel-reserved boot regions from EFI page tables Ard Biesheuvel
2026-04-23 15:20 ` [PATCH v3 07/17] x86/efi: Drop EFI_MEMORY_RUNTIME check from __ioremap_check_other() Ard Biesheuvel
2026-04-23 15:20 ` [PATCH v3 08/17] x86/efi: Allow ranges_to_free array to grow beyond initial size Ard Biesheuvel
2026-04-23 15:20 ` [PATCH v3 09/17] x86/efi: Intersect ranges_to_free with MEMBLOCK_RSRV_KERN regions Ard Biesheuvel
2026-04-23 15:20 ` [PATCH v3 10/17] x86/efi: Do not rely on EFI_MEMORY_RUNTIME bit and avoid entry splitting Ard Biesheuvel
2026-04-23 15:20 ` [PATCH v3 11/17] efi: Use nr_map not map_end to find the last valid memory map entry Ard Biesheuvel
2026-04-23 15:20 ` [PATCH v3 12/17] x86/efi: Clean the memory map using iterator and filter API Ard Biesheuvel
2026-04-23 15:20 ` [PATCH v3 13/17] x86/efi: Update the runtime map in place Ard Biesheuvel
2026-04-23 15:20 ` [PATCH v3 14/17] x86/efi: Reuse memory map instead of reallocating it Ard Biesheuvel
2026-04-23 15:20 ` [PATCH v3 15/17] x86/efi: Merge two traversals of the memory map when freeing boot regions Ard Biesheuvel
2026-04-23 15:20 ` Ard Biesheuvel [this message]
2026-04-23 15:20 ` [PATCH v3 17/17] x86/efi: Drop kexec quirk for the EFI memory attributes table Ard Biesheuvel

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260423152024.1098465-35-ardb+git@google.com \
    --to=ardb+git@google.com \
    --cc=ardb@kernel.org \
    --cc=benh@kernel.crashing.org \
    --cc=gourry@gourry.net \
    --cc=linux-efi@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=rppt@kernel.org \
    --cc=ruirui.yang@linux.dev \
    --cc=x86@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox