Received-SPF: pass (google.com: domain of linux-kernel+bounces-5461-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
From: Vegard Nossum <vegard.nossum@oracle.com>
To: Thomas Gleixner <tglx@linutronix.de>, Ingo Molnar <mingo@redhat.com>,
        Borislav Petkov <bp@alien8.de>,
        Dave Hansen <dave.hansen@linux.intel.com>, x86@kernel.org
Cc: "H. Peter Anvin" <hpa@zytor.com>, linux-kernel@vger.kernel.org,
        Brian Gerst <brgerst@gmail.com>, Peter Zijlstra <peterz@infradead.org>,
        Linus Torvalds <torvalds@linuxfoundation.org>,
        Vegard Nossum <vegard.nossum@oracle.com>
Subject: [PATCH 5/5] x86: add DB flag to 32-bit percpu GDT entry
Date: Tue, 19 Dec 2023 16:12:00 +0100
Message-Id: <20231219151200.2878271-6-vegard.nossum@oracle.com>
In-Reply-To: <20231219151200.2878271-1-vegard.nossum@oracle.com>
References: <20231219151200.2878271-1-vegard.nossum@oracle.com>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit

The D/B size flag for the 32-bit percpu GDT entry was not set.

The Intel manual (vol 3, section 3.4.5) only specifies the meaning of
this flag for three cases:

1) code segments used for %cs -- doesn't apply here
2) stack segments used for %ss -- doesn't apply
3) expand-down data segments -- but we don't have the expand-down flag
   set, so it also doesn't apply here

The flag likely doesn't do anything here, although the manual does also
say: "This flag should always be set to 1 for 32-bit code and data
segments [...]" so we should probably do it anyway.

Link: https://lore.kernel.org/all/CAHk-=wib5XLebuEra7y2YH96wxdk=8vJnA8XoVq0FExpzVvN=Q@mail.gmail.com/
Signed-off-by: Vegard Nossum <vegard.nossum@oracle.com>
---
 arch/x86/kernel/setup_percpu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c
index f2583de97a64..b30d6e180df7 100644
--- a/arch/x86/kernel/setup_percpu.c
+++ b/arch/x86/kernel/setup_percpu.c
@@ -106,7 +106,7 @@ void __init pcpu_populate_pte(unsigned long addr)
 static inline void setup_percpu_segment(int cpu)
 {
 #ifdef CONFIG_X86_32
-	struct desc_struct d = GDT_ENTRY_INIT(DESC_DATA32 & ~_DESC_DB,
+	struct desc_struct d = GDT_ENTRY_INIT(DESC_DATA32,
 					      per_cpu_offset(cpu), 0xFFFFF);
 
 	write_gdt_entry(get_cpu_gdt_rw(cpu), GDT_ENTRY_PERCPU, &d, DESCTYPE_S);
-- 
2.34.1