From: "Ard Biesheuvel" <ardb@kernel.org>
To: devel@edk2.groups.io
Cc: leif@nuviainc.com, sami.mujawar@arm.com,
Ard Biesheuvel <ardb@kernel.org>,
Alexander Graf <agraf@csgraf.de>
Subject: [PATCH] ArmVirtPkg/ArmVirtMemoryInitPeiLib: avoid redundant cache invalidation
Date: Sat, 29 Jan 2022 16:13:46 +0100 [thread overview]
Message-ID: <20220129151346.4120964-1-ardb@kernel.org> (raw)
Alex reports that the cache invalidation performed by
ArmVirtMemoryInitPeiLib takes a non-negligible amount of time at boot.
This cache invalidation used to be necessary to avoid inconsistencies
between the CPU's cached and uncached views of the permanent PEI memory
region, given that the PEI phase is where the MMU gets enabled.
The only allocations done from permanent PEI memory with the MMU off are
pages used for page tables, and since commit 748fea6279ef
("ArmPkg/ArmMmuLib AARCH64: invalidate page tables before populating
them"), each of those is invalidated in the caches explicitly, for
reasons described in the patch's commit log. All other allocations done
in PEI are either from temporary PEI memory, which includes the stack,
or from permanent PEI memory but after the MMU has been enabled.
This means that the cache invalidation in ArmVirtMemoryInitPeiLib is no
longer necessary, and can simply be dropped.
Cc: Alexander Graf <agraf@csgraf.de>
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
---
.../ArmVirtMemoryInitPeiLib/ArmVirtMemoryInitPeiLib.c | 9 ---------
1 file changed, 9 deletions(-)
diff --git a/ArmVirtPkg/Library/ArmVirtMemoryInitPeiLib/ArmVirtMemoryInitPeiLib.c b/ArmVirtPkg/Library/ArmVirtMemoryInitPeiLib/ArmVirtMemoryInitPeiLib.c
index 022e13e762b6..98d90ad4200d 100644
--- a/ArmVirtPkg/Library/ArmVirtMemoryInitPeiLib/ArmVirtMemoryInitPeiLib.c
+++ b/ArmVirtPkg/Library/ArmVirtMemoryInitPeiLib/ArmVirtMemoryInitPeiLib.c
@@ -91,15 +91,6 @@ MemoryPeim (
);
}
- //
- // When running under virtualization, the PI/UEFI memory region may be
- // clean but not invalidated in system caches or in lower level caches
- // on other CPUs. So invalidate the region by virtual address, to ensure
- // that the contents we put there with the caches and MMU off will still
- // be visible after turning them on.
- //
- InvalidateDataCacheRange ((VOID *)(UINTN)UefiMemoryBase, UefiMemorySize);
-
// Build Memory Allocation Hob
InitMmu ();
--
2.30.2
next reply other threads:[~2022-01-29 15:13 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-01-29 15:13 Ard Biesheuvel [this message]
2022-01-31 12:22 ` [PATCH] ArmVirtPkg/ArmVirtMemoryInitPeiLib: avoid redundant cache invalidation Leif Lindholm
2022-01-31 13:21 ` Alexander Graf
2022-02-25 16:48 ` 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-list 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=20220129151346.4120964-1-ardb@kernel.org \
--to=devel@edk2.groups.io \
/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