Re: [tip:x86/apic] x86/idt: Move IST stack based traps to table init

From: Andrei Vagin
Date: Fri Sep 01 2017 - 02:28:44 EST


Hi Thomas,

We run CRIU tests for linux-next and today they detected an issue. I've
bisected this problem and it looks like a problem is in this patch.

[root@fc24 criu]# python ./test/zdtm.py run -t zdtm/static/env00
=== Run 1/1 ================ zdtm/static/env00

========================== Run zdtm/static/env00 in h ==========================
Start test
./env00 --pidfile=env00.pid --outfile=env00.out --envname=ENV_00_TEST
Run criu dump
=[log]=> dump/zdtm/static/env00/36/1/dump.log
------------------------ grep Error ------------------------
(00.004013) Dump private signals of 36
(00.004022) Dump shared signals of 36
(00.004036) Parasite syscall_ip at 0x400000
(00.004097) ** delivering signal 11 si_code=128
(00.004103) Error (compel/src/lib/infect.c:528): Unexpected 36 task interruption, aborting
(00.004118) Error (criu/cr-dump.c:1371): Can't infect (pid: 36) with parasite
(00.004208) Unlock network
(00.004218) Unfreezing tasks into 1
(00.004223) Unseizing 36 into 1
(00.004244) Error (criu/cr-dump.c:1807): Dumping FAILED.
------------------------ ERROR OVER ------------------------
################### Test zdtm/static/env00 FAIL at CRIU dump ###################
Send the 9 signal to 36
Wait for zdtm/static/env00(36) to die for 0.100000
##################################### FAIL #####################################

Here is a line in a code where we get this unexpected error:
https://github.com/xemul/criu/blob/criu-dev/compel/src/lib/infect.c#L735

At this moment criu tries to execute the memfd_create syscall in a
context of another task with help of ptrace.

Here is a link to a test job:
https://travis-ci.org/avagin/linux/builds/270623449

Thanks,
Andrei

On Tue, Aug 29, 2017 at 04:18:51AM -0700, tip-bot for Jacob Shin wrote:
> Commit-ID: 90f6225fba0c732f3f5f9f5e265bdefa021ff12d
> Gitweb: http://git.kernel.org/tip/90f6225fba0c732f3f5f9f5e265bdefa021ff12d
> Author: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
> AuthorDate: Mon, 28 Aug 2017 08:47:52 +0200
> Committer: Ingo Molnar <mingo@xxxxxxxxxx>
> CommitDate: Tue, 29 Aug 2017 12:07:27 +0200
>
> x86/idt: Move IST stack based traps to table init
>
> Initialize the IST based traps via a table.
>
> Signed-off-by: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
> Cc: Andy Lutomirski <luto@xxxxxxxxxx>
> Cc: Borislav Petkov <bp@xxxxxxxxx>
> Cc: Brian Gerst <brgerst@xxxxxxxxx>
> Cc: Denys Vlasenko <dvlasenk@xxxxxxxxxx>
> Cc: H. Peter Anvin <hpa@xxxxxxxxx>
> Cc: Josh Poimboeuf <jpoimboe@xxxxxxxxxx>
> Cc: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
> Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> Cc: Steven Rostedt <rostedt@xxxxxxxxxxx>
> Link: http://lkml.kernel.org/r/20170828064959.091328949@xxxxxxxxxxxxx
> Signed-off-by: Ingo Molnar <mingo@xxxxxxxxxx>
> ---
> arch/x86/include/asm/desc.h | 2 ++
> arch/x86/kernel/idt.c | 22 ++++++++++++++++++++++
> arch/x86/kernel/traps.c | 9 +--------
> 3 files changed, 25 insertions(+), 8 deletions(-)
>
> diff --git a/arch/x86/include/asm/desc.h b/arch/x86/include/asm/desc.h
> index 930acd5..e624527 100644
> --- a/arch/x86/include/asm/desc.h
> +++ b/arch/x86/include/asm/desc.h
> @@ -509,9 +509,11 @@ extern void idt_setup_early_traps(void);
>
> #ifdef CONFIG_X86_64
> extern void idt_setup_early_pf(void);
> +extern void idt_setup_ist_traps(void);
> extern void idt_setup_debugidt_traps(void);
> #else
> static inline void idt_setup_early_pf(void) { }
> +static inline void idt_setup_ist_traps(void) { }
> static inline void idt_setup_debugidt_traps(void) { }
> #endif
>
> diff --git a/arch/x86/kernel/idt.c b/arch/x86/kernel/idt.c
> index f5281b8..a6326fd 100644
> --- a/arch/x86/kernel/idt.c
> +++ b/arch/x86/kernel/idt.c
> @@ -92,6 +92,20 @@ struct desc_ptr idt_descr __ro_after_init = {
> gate_desc debug_idt_table[IDT_ENTRIES] __page_aligned_bss;
>
> /*
> + * The exceptions which use Interrupt stacks. They are setup after
> + * cpu_init() when the TSS has been initialized.
> + */
> +static const __initdata struct idt_data ist_idts[] = {
> + ISTG(X86_TRAP_DB, debug, DEBUG_STACK),
> + ISTG(X86_TRAP_NMI, nmi, NMI_STACK),
> + ISTG(X86_TRAP_BP, int3, DEBUG_STACK),
> + ISTG(X86_TRAP_DF, double_fault, DOUBLEFAULT_STACK),
> +#ifdef CONFIG_X86_MCE
> + ISTG(X86_TRAP_MC, &machine_check, MCE_STACK),
> +#endif
> +};
> +
> +/*
> * Override for the debug_idt. Same as the default, but with interrupt
> * stack set to DEFAULT_STACK (0). Required for NMI trap handling.
> */
> @@ -158,6 +172,14 @@ void __init idt_setup_early_pf(void)
> }
>
> /**
> + * idt_setup_ist_traps - Initialize the idt table with traps using IST
> + */
> +void __init idt_setup_ist_traps(void)
> +{
> + idt_setup_from_table(idt_table, ist_idts, ARRAY_SIZE(ist_idts));
> +}
> +
> +/**
> * idt_setup_debugidt_traps - Initialize the debug idt table with debug traps
> */
> void __init idt_setup_debugidt_traps(void)
> diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c
> index 1492bf5..293f5bd 100644
> --- a/arch/x86/kernel/traps.c
> +++ b/arch/x86/kernel/traps.c
> @@ -979,14 +979,7 @@ void __init trap_init(void)
> */
> cpu_init();
>
> - /*
> - * X86_TRAP_DB and X86_TRAP_BP have been set
> - * in early_trap_init(). However, ITS works only after
> - * cpu_init() loads TSS. See comments in early_trap_init().
> - */
> - set_intr_gate_ist(X86_TRAP_DB, &debug, DEBUG_STACK);
> - /* int3 can be called from all */
> - set_system_intr_gate_ist(X86_TRAP_BP, &int3, DEBUG_STACK);
> + idt_setup_ist_traps();
>
> x86_init.irqs.trap_init();
>