From: Andy Lutomirski <luto@kernel.org>
To: x86@kernel.org
Cc: LKML <linux-kernel@vger.kernel.org>,
"Krzysztof Mazur" <krzysiek@podlesie.net>,
"Krzysztof Olędzki" <ole@ans.pl>, "Arnd Bergmann" <arnd@arndb.de>,
"Andy Lutomirski" <luto@kernel.org>
Subject: [PATCH 3/4] x86/fpu: Make the EFI FPU calling convention explicit
Date: Sun, 17 Jan 2021 22:20:40 -0800 [thread overview]
Message-ID: <49aeb632f155569c75ea3cd78e9028cf7ba776fe.1610950681.git.luto@kernel.org> (raw)
In-Reply-To: <cover.1610950681.git.luto@kernel.org>
EFI uses kernel_fpu_begin() to conform to the UEFI calling convention.
This specifically requires initializing FCW, whereas no sane 64-bit kernel
code should use legacy 387 operations that reference FCW.
Add KFPU_EFI to make this self-documenting, and use it in the EFI code.
This should enable us to safely change the default semantics of
kernel_fpu_begin() to stop initializing FCW on 64-bit kernels.
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andy Lutomirski <luto@kernel.org>
---
arch/x86/include/asm/efi.h | 4 ++--
arch/x86/include/asm/fpu/api.h | 7 +++++++
arch/x86/platform/efi/efi_64.c | 2 +-
3 files changed, 10 insertions(+), 3 deletions(-)
diff --git a/arch/x86/include/asm/efi.h b/arch/x86/include/asm/efi.h
index bc9758ef292e..c60be69a5c82 100644
--- a/arch/x86/include/asm/efi.h
+++ b/arch/x86/include/asm/efi.h
@@ -71,7 +71,7 @@ extern unsigned long efi_fw_vendor, efi_config_table;
#ifdef CONFIG_X86_32
#define arch_efi_call_virt_setup() \
({ \
- kernel_fpu_begin(); \
+ kernel_fpu_begin_mask(KFPU_EFI); \
firmware_restrict_branch_speculation_start(); \
})
@@ -107,7 +107,7 @@ struct efi_scratch {
#define arch_efi_call_virt_setup() \
({ \
efi_sync_low_kernel_mappings(); \
- kernel_fpu_begin(); \
+ kernel_fpu_begin_mask(KFPU_EFI); \
firmware_restrict_branch_speculation_start(); \
efi_switch_mm(&efi_mm); \
})
diff --git a/arch/x86/include/asm/fpu/api.h b/arch/x86/include/asm/fpu/api.h
index 133907a200ef..e95a06845443 100644
--- a/arch/x86/include/asm/fpu/api.h
+++ b/arch/x86/include/asm/fpu/api.h
@@ -25,6 +25,13 @@
#define KFPU_XYZMM _BITUL(1) /* MXCSR will be initialized */
#define KFPU_MMX 0 /* nothing gets initialized */
+/*
+ * The UEFI calling convention (UEFI spec 2.3.2 and 2.3.4) requires
+ * that FCW (32-bit and 64-bit) and MXCSR (64-bit) must be initialized
+ * prior to calling UEFI code.
+ */
+#define KFPU_EFI (KFPU_387 | KFPU_XYZMM)
+
extern void kernel_fpu_begin_mask(unsigned int kfpu_mask);
extern void kernel_fpu_end(void);
extern bool irq_fpu_usable(void);
diff --git a/arch/x86/platform/efi/efi_64.c b/arch/x86/platform/efi/efi_64.c
index 8f5759df7776..c304c8da862b 100644
--- a/arch/x86/platform/efi/efi_64.c
+++ b/arch/x86/platform/efi/efi_64.c
@@ -848,7 +848,7 @@ efi_set_virtual_address_map(unsigned long memory_map_size,
virtual_map);
efi_switch_mm(&efi_mm);
- kernel_fpu_begin();
+ kernel_fpu_begin_mask(KFPU_EFI);
/* Disable interrupts around EFI calls: */
local_irq_save(flags);
--
2.29.2
next prev parent reply other threads:[~2021-01-18 6:22 UTC|newest]
Thread overview: 10+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-01-18 6:20 [PATCH 0/4] x86/fpu: Reduce unnecessary FNINIT and MXCSR usage Andy Lutomirski
2021-01-18 6:20 ` [PATCH 1/4] x86/fpu: Add kernel_fpu_begin_mask() to selectively initialize state Andy Lutomirski
2021-01-18 12:21 ` Peter Zijlstra
2021-01-18 14:03 ` David Laight
2021-01-19 11:08 ` Borislav Petkov
2021-01-18 6:20 ` [PATCH 2/4] x86/mmx: Use KFPU_MMX for MMX string operations Andy Lutomirski
2021-01-18 6:20 ` Andy Lutomirski [this message]
2021-01-18 12:13 ` [PATCH 3/4] x86/fpu: Make the EFI FPU calling convention explicit Borislav Petkov
2021-01-18 6:20 ` [PATCH 4/4] x86/fpu/64: Don't FNINIT in kernel_fpu_begin() Andy Lutomirski
2021-01-18 8:25 ` [PATCH 0/4] x86/fpu: Reduce unnecessary FNINIT and MXCSR usage Krzysztof Olędzki
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=49aeb632f155569c75ea3cd78e9028cf7ba776fe.1610950681.git.luto@kernel.org \
--to=luto@kernel.org \
--cc=arnd@arndb.de \
--cc=krzysiek@podlesie.net \
--cc=linux-kernel@vger.kernel.org \
--cc=ole@ans.pl \
--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;
as well as URLs for NNTP newsgroup(s).