linux-kernel - [PATCH 6/6] pre states for x86

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

Message-Id: <20210201103835.1043254-6-dwmw@amazon.co.uk>
Date:   Mon,  1 Feb 2021 10:38:35 +0000
From:   David Woodhouse <dwmw@...zon.co.uk>
To:     Thomas Gleixner <tglx@...utronix.de>
Cc:     Andy Lutomirski <luto@...nel.org>,
        "shenkai (D)" <shenkai8@...wei.com>, mimoja@...zon.com,
        LKML <linux-kernel@...r.kernel.org>,
        Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
        X86 ML <x86@...nel.org>, "H . Peter Anvin" <hpa@...or.com>,
        hewenliang4@...wei.com, hushiyuan@...wei.com,
        luolongjun@...wei.com, hejingxian@...wei.com
Subject: [PATCH 6/6] pre states for x86

Utterly broken because the bringup uses global initial_stack and initial_gs
variables, and the TSC sync is similarly hosed (I should probably do one
at a time).

The bringup is going to be the most fun to fix, because the AP coming up
doesn't actually have a lot that it *can* use to disambiguate itself from
the others. Starting them at different locations is mostly a non-starter
as we can only specify a page address under 1MiB and there are only 256
of those even if we could allocate them *all* to start different CPUs.
I think we need to get them to find their own logical CPU# from their
APICID, perhaps by trawling the per_cpu x86_cpu_to_apicid or some other
means, then find their initial stack / %gs from that.

We also need to work out if we can eliminate the real-mode stack for the
trampoline, or do some locking if we really must share it perhaps.
---
 arch/x86/kernel/smpboot.c | 24 ++++++++++++++++--------
 1 file changed, 16 insertions(+), 8 deletions(-)

diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index 649b8236309b..03f63027fdad 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -57,6 +57,7 @@
 #include <linux/pgtable.h>
 #include <linux/overflow.h>
 #include <linux/syscore_ops.h>
+#include <linux/smpboot.h>
 
 #include <asm/acpi.h>
 #include <asm/desc.h>
@@ -1217,14 +1218,6 @@ int native_cpu_up(unsigned int cpu, struct task_struct *tidle)
 {
 	int ret;
 
-	ret = do_cpu_up(cpu, tidle);
-	if (ret)
-		return ret;
-
-	ret = do_wait_cpu_initialized(cpu);
-	if (ret)
-		return ret;
-
 	ret = do_wait_cpu_callin(cpu);
 	if (ret)
 		return ret;
@@ -1241,6 +1234,16 @@ int native_cpu_up(unsigned int cpu, struct task_struct *tidle)
 	return ret;
 }
 
+int native_cpu_kick(unsigned int cpu)
+{
+	return do_cpu_up(cpu, idle_thread_get(cpu));
+}
+
+int native_cpu_wait_init(unsigned int cpu)
+{
+	return do_wait_cpu_initialized(cpu);
+}
+
 /**
  * arch_disable_smp_support() - disables SMP support for x86 at runtime
  */
@@ -1412,6 +1415,11 @@ void __init native_smp_prepare_cpus(unsigned int max_cpus)
 	smp_quirk_init_udelay();
 
 	speculative_store_bypass_ht_init();
+
+	cpuhp_setup_state_nocalls(CPUHP_BP_PARALLEL_DYN, "x86/cpu:kick",
+				  native_cpu_kick, NULL);
+	cpuhp_setup_state_nocalls(CPUHP_BP_PARALLEL_DYN, "x86/cpu:wait-init",
+				  native_cpu_wait_init, NULL);
 }
 
 void arch_thaw_secondary_cpus_begin(void)
-- 
2.29.2