From bc696ca05f5a8927329ec276a892341e006b00ba Mon Sep 17 00:00:00 2001 From: Borislav Petkov Date: Tue, 26 Jan 2016 22:12:05 +0100 Subject: x86/cpufeature: Replace the old static_cpu_has() with safe variant So the old one didn't work properly before alternatives had run. And it was supposed to provide an optimized JMP because the assumption was that the offset it is jumping to is within a signed byte and thus a two-byte JMP. So I did an x86_64 allyesconfig build and dumped all possible sites where static_cpu_has() was used. The optimization amounted to all in all 12(!) places where static_cpu_has() had generated a 2-byte JMP. Which has saved us a whopping 36 bytes! This clearly is not worth the trouble so we can remove it. The only place where the optimization might count - in __switch_to() - we will handle differently. But that's not subject of this patch. Signed-off-by: Borislav Petkov Cc: Andy Lutomirski Cc: Borislav Petkov Cc: Brian Gerst Cc: Denys Vlasenko Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1453842730-28463-6-git-send-email-bp@alien8.de Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/common.c | 12 ++---------- 1 file changed, 2 insertions(+), 10 deletions(-) (limited to 'arch/x86/kernel/cpu/common.c') diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index 37830de8f60a..ee499817f3f5 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -1475,19 +1475,11 @@ void cpu_init(void) } #endif -#ifdef CONFIG_X86_DEBUG_STATIC_CPU_HAS -void warn_pre_alternatives(void) -{ - WARN(1, "You're using static_cpu_has before alternatives have run!\n"); -} -EXPORT_SYMBOL_GPL(warn_pre_alternatives); -#endif - -inline bool __static_cpu_has_safe(u16 bit) +inline bool __static_cpu_has(u16 bit) { return boot_cpu_has(bit); } -EXPORT_SYMBOL_GPL(__static_cpu_has_safe); +EXPORT_SYMBOL_GPL(__static_cpu_has); static void bsp_resume(void) { -- cgit v1.2.3 From 2476f2fa20568bd5d9e09cd35bcd73e99a6f4cc6 Mon Sep 17 00:00:00 2001 From: Brian Gerst Date: Wed, 27 Jan 2016 09:45:25 +0100 Subject: x86/alternatives: Discard dynamic check after init Move the code to do the dynamic check to the altinstr_aux section so that it is discarded after alternatives have run and a static branch has been chosen. This way we're changing the dynamic branch from C code to assembly, which makes it *substantially* smaller while avoiding a completely unnecessary call to an out of line function. Signed-off-by: Brian Gerst [ Changed it to do TESTB, as hpa suggested. ] Signed-off-by: Borislav Petkov Cc: Andrew Morton Cc: Andy Lutomirski Cc: Andy Lutomirski Cc: Boris Ostrovsky Cc: Borislav Petkov Cc: Dave Young Cc: Denys Vlasenko Cc: H. Peter Anvin Cc: Kristen Carlson Accardi Cc: Laura Abbott Cc: Linus Torvalds Cc: Peter Zijlstra (Intel) Cc: Peter Zijlstra Cc: Prarit Bhargava Cc: Ross Zwisler Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1452972124-7380-1-git-send-email-brgerst@gmail.com Link: http://lkml.kernel.org/r/20160127084525.GC30712@pd.tnic Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/common.c | 6 ------ 1 file changed, 6 deletions(-) (limited to 'arch/x86/kernel/cpu/common.c') diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index ee499817f3f5..079d83fc6488 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -1475,12 +1475,6 @@ void cpu_init(void) } #endif -inline bool __static_cpu_has(u16 bit) -{ - return boot_cpu_has(bit); -} -EXPORT_SYMBOL_GPL(__static_cpu_has); - static void bsp_resume(void) { if (this_cpu->c_bsp_resume) -- cgit v1.2.3 From 58a5aac5331388a175a42b6ed2154f0559cefb21 Mon Sep 17 00:00:00 2001 From: Andy Lutomirski Date: Mon, 29 Feb 2016 15:50:19 -0800 Subject: x86/entry/32: Introduce and use X86_BUG_ESPFIX instead of paravirt_enabled x86_64 has very clean espfix handling on paravirt: espfix64 is set up in native_iret, so paravirt systems that override iret bypass espfix64 automatically. This is robust and straightforward. x86_32 is messier. espfix is set up before the IRET paravirt patch point, so it can't be directly conditionalized on whether we use native_iret. We also can't easily move it into native_iret without regressing performance due to a bizarre consideration. Specifically, on 64-bit kernels, the logic is: if (regs->ss & 0x4) setup_espfix; On 32-bit kernels, the logic is: if ((regs->ss & 0x4) && (regs->cs & 0x3) == 3 && (regs->flags & X86_EFLAGS_VM) == 0) setup_espfix; The performance of setup_espfix itself is essentially irrelevant, but the comparison happens on every IRET so its performance matters. On x86_64, there's no need for any registers except flags to implement the comparison, so we fold the whole thing into native_iret. On x86_32, we don't do that because we need a free register to implement the comparison efficiently. We therefore do espfix setup before restoring registers on x86_32. This patch gets rid of the explicit paravirt_enabled check by introducing X86_BUG_ESPFIX on 32-bit systems and using an ALTERNATIVE to skip espfix on paravirt systems where iret != native_iret. This is also messy, but it's at least in line with other things we do. This improves espfix performance by removing a branch, but no one cares. More importantly, it removes a paravirt_enabled user, which is good because paravirt_enabled is ill-defined and is going away. Signed-off-by: Andy Lutomirski Reviewed-by: Borislav Petkov Cc: Andrew Cooper Cc: Andy Lutomirski Cc: Borislav Petkov Cc: Brian Gerst Cc: Denys Vlasenko Cc: H. Peter Anvin Cc: Linus Torvalds Cc: Luis R. Rodriguez Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: boris.ostrovsky@oracle.com Cc: david.vrabel@citrix.com Cc: konrad.wilk@oracle.com Cc: lguest@lists.ozlabs.org Cc: xen-devel@lists.xensource.com Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/common.c | 25 +++++++++++++++++++++++++ 1 file changed, 25 insertions(+) (limited to 'arch/x86/kernel/cpu/common.c') diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index 079d83fc6488..d8337f34b5f4 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -802,6 +802,31 @@ static void detect_nopl(struct cpuinfo_x86 *c) clear_cpu_cap(c, X86_FEATURE_NOPL); #else set_cpu_cap(c, X86_FEATURE_NOPL); +#endif + + /* + * ESPFIX is a strange bug. All real CPUs have it. Paravirt + * systems that run Linux at CPL > 0 may or may not have the + * issue, but, even if they have the issue, there's absolutely + * nothing we can do about it because we can't use the real IRET + * instruction. + * + * NB: For the time being, only 32-bit kernels support + * X86_BUG_ESPFIX as such. 64-bit kernels directly choose + * whether to apply espfix using paravirt hooks. If any + * non-paravirt system ever shows up that does *not* have the + * ESPFIX issue, we can change this. + */ +#ifdef CONFIG_X86_32 +#ifdef CONFIG_PARAVIRT + do { + extern void native_iret(void); + if (pv_cpu_ops.iret == native_iret) + set_cpu_bug(c, X86_BUG_ESPFIX); + } while (0); +#else + set_cpu_bug(c, X86_BUG_ESPFIX); +#endif #endif } -- cgit v1.2.3