linux-kernel - Re: [RFC PATCH 5/6] x86/fpu: Fix CPUID-less FPU detection

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20161207094229.gxqatdjulfvnz7eo@pd.tnic>
Date:   Wed, 7 Dec 2016 10:42:29 +0100
From:   Borislav Petkov <bp@...en8.de>
To:     Andy Lutomirski <luto@...capital.net>
Cc:     Andy Lutomirski <luto@...nel.org>, X86 ML <x86@...nel.org>,
        One Thousand Gnomes <gnomes@...rguk.ukuu.org.uk>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        Brian Gerst <brgerst@...il.com>,
        Matthew Whitehead <tedheadster@...il.com>
Subject: Re: [RFC PATCH 5/6] x86/fpu: Fix CPUID-less FPU detection

On Tue, Dec 06, 2016 at 09:52:11AM -0800, Andy Lutomirski wrote:
> How?  There's that bit at the bottom to worry about.

Even better: carve it out into a separate function. It was begging for
it already. Diff ontop of yours:

---
Index: b/arch/x86/kernel/fpu/init.c
===================================================================
--- a/arch/x86/kernel/fpu/init.c	2016-12-07 10:39:57.643083996 +0100
+++ b/arch/x86/kernel/fpu/init.c	2016-12-07 10:39:45.807083690 +0100
@@ -49,31 +49,37 @@ void fpu__init_cpu(void)
 }
 
 /*
- * The earliest FPU detection code.
+ * Try to execute an actual sequence of FPU instructions. Our last resort for
+ * FPU detection in case the respective CPUID bit is not set or we don't
+ * support CPUID at all.
  */
-static void fpu__init_system_early_generic(struct cpuinfo_x86 *c)
+static bool __fpu__init_poke_fpu(void)
 {
-	if (!boot_cpu_has(X86_FEATURE_CPUID) &&
-	    !test_bit(X86_FEATURE_FPU, (unsigned long *)cpu_caps_cleared)) {
-		/*
-		 * Set the X86_FEATURE_FPU CPU-capability bit based on
-		 * trying to execute an actual sequence of FPU instructions:
-		 */
+	unsigned long cr0;
+	u16 fsw, fcw;
 
-		unsigned long cr0;
-		u16 fsw, fcw;
+	fsw = fcw = 0xffff;
 
-		fsw = fcw = 0xffff;
+	cr0 = read_cr0();
+	cr0 &= ~(X86_CR0_TS | X86_CR0_EM);
+	write_cr0(cr0);
 
-		cr0 = read_cr0();
-		cr0 &= ~(X86_CR0_TS | X86_CR0_EM);
-		write_cr0(cr0);
+	asm volatile("fninit ; fnstsw %0 ; fnstcw %1"
+		     : "+m" (fsw), "+m" (fcw));
 
-		asm volatile("fninit ; fnstsw %0 ; fnstcw %1"
-			     : "+m" (fsw), "+m" (fcw));
-		pr_info("x86/fpu: FSW=0x%04hx FCW=0x%04hx\n", fsw, fcw);
+	pr_info("x86/fpu: FSW=0x%04hx FCW=0x%04hx\n", fsw, fcw);
 
-		if (fsw == 0 && (fcw & 0x103f) == 0x003f)
+	return fsw == 0 && (fcw & 0x103f) == 0x003f;
+}
+
+/*
+ * The earliest FPU detection code.
+ */
+static void fpu__init_system_early_generic(struct cpuinfo_x86 *c)
+{
+	if (!boot_cpu_has(X86_FEATURE_CPUID) &&
+	    !test_bit(X86_FEATURE_FPU, (unsigned long *)cpu_caps_cleared)) {
+		if (__fpu__init_poke_fpu())
 			setup_force_cpu_cap(X86_FEATURE_FPU);
 		else
 			setup_clear_cpu_cap(X86_FEATURE_FPU);

> I kind of line the idea of being able to ask users of these ancient
> CPUs to just send in their logs.

Prudent :)

-- 
Regards/Gruss,
    Boris.

Good mailing practices for 400: avoid top-posting and trim the reply.