linux-kernel - RE: [PATCH] Debug: Add cpu registers dump in case of kernel panic

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

Message-ID: <EEBA739CCF11FE49B73E1FB4690F5EE649E29D94@SHSMSX101.ccr.corp.intel.com>
Date:   Thu, 5 Jul 2018 06:52:09 +0000
From:   "Tian, Baofeng" <baofeng.tian@...el.com>
To:     "'tglx@...utronix.de'" <tglx@...utronix.de>,
        "'mingo@...hat.com'" <mingo@...hat.com>,
        "'hpa@...or.com'" <hpa@...or.com>,
        "'x86@...nel.org'" <x86@...nel.org>,
        "'rostedt@...dmis.org'" <rostedt@...dmis.org>,
        "'linux-kernel@...r.kernel.org'" <linux-kernel@...r.kernel.org>
Subject: RE: [PATCH] Debug: Add cpu registers dump in case of kernel panic

Resubmit again for review, this patch save register and used for RAM dump.

Ramdump is an debug feature used to analysis complex panic/hang issues, it will recover all the data after pull the RAM to host and use an host tool to decode and analysis the RAM.

Thanks
Tim

From: Tian, Baofeng 
Sent: Monday, July 2, 2018 8:46 AM
To: tglx@...utronix.de; mingo@...hat.com; hpa@...or.com; x86@...nel.org; rostedt@...dmis.org; linux-kernel@...r.kernel.org
Subject: [PATCH] Debug: Add cpu registers dump in case of kernel panic

From: "Tian, Baofeng" <baofeng.tian@...el.com>
Subject: [PATCH] Debug: Add cpu registers dump in case of kernel panic

When kernel panic, the value of cpu registers are needed
for analyzing, through whole ramdump to restore the crash
point.

We store CPU registers to RAM before cpu stopped, then get
them with whole ram dump after warm reset.Those registers
are used in analysis of dumped RAM image.

Signed-off-by: Baofeng, Tian <baofeng.tian@...el.com>
Signed-off-by: Emmanuel Berthier <emmanuel.berthier@...el.com>
---
 arch/x86/kernel/smp.c | 19 ++++++++++++++++++-
 1 file changed, 18 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kernel/smp.c b/arch/x86/kernel/smp.c
index 5c574df..484831d 100644
--- a/arch/x86/kernel/smp.c
+++ b/arch/x86/kernel/smp.c
@@ -117,6 +117,20 @@
 static atomic_t stopping_cpu = ATOMIC_INIT(-1);
 static bool smp_no_nmi_ipi = false;
 
+static DEFINE_PER_CPU(struct pt_regs, cpu_regs);
+
+/* Store regs of this CPU for RAM dump decoding help */
+static inline void store_regs(struct pt_regs *regs)
+{
+     struct pt_regs *print_regs;
+
+     print_regs = &get_cpu_var(cpu_regs);
+     crash_setup_regs(print_regs, regs);
+
+     /* Flush CPU cache */
+     wbinvd();
+}
+
 /*
  * this function sends a 'reschedule' IPI to another CPU.
  * it goes straight through and wastes no time serializing
@@ -163,6 +177,7 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
      if (raw_smp_processor_id() == atomic_read(&stopping_cpu))
            return NMI_HANDLED;
 
+     store_regs(regs);
      cpu_emergency_vmxoff();
      stop_this_cpu(NULL);
 
@@ -173,9 +188,10 @@ static int smp_stop_nmi_callback(unsigned int val, struct pt_regs *regs)
  * this function calls the 'stop' function on all other CPUs in the system.
  */
 
-asmlinkage __visible void smp_reboot_interrupt(void)
+__visible void smp_reboot_interrupt(struct pt_regs *regs)
 {
      ipi_entering_ack_irq();
+     store_regs(regs);
      cpu_emergency_vmxoff();
      stop_this_cpu(NULL);
      irq_exit();
@@ -247,6 +263,7 @@ static void native_stop_other_cpus(int wait)
      }
 
 finish:
+     store_regs(NULL);
      local_irq_save(flags);
      disable_local_APIC();
      mcheck_cpu_clear(this_cpu_ptr(&cpu_info));
-- 
2.7.4