From: Juergen Gross <jgross@suse.com>
To: Thomas Gleixner <tglx@linutronix.de>,
Andrew Cooper <Andrew.Cooper3@citrix.com>,
LKML <linux-kernel@vger.kernel.org>
Cc: "x86@kernel.org" <x86@kernel.org>,
Linus Torvalds <torvalds@linux-foundation.org>,
Tim Chen <tim.c.chen@linux.intel.com>,
Josh Poimboeuf <jpoimboe@kernel.org>,
Pawan Gupta <pawan.kumar.gupta@linux.intel.com>,
Johannes Wikner <kwikner@ethz.ch>,
Alyssa Milburn <alyssa.milburn@linux.intel.com>,
Jann Horn <jannh@google.com>, "H.J. Lu" <hjl.tools@gmail.com>,
Joao Moreira <joao.moreira@intel.com>,
Joseph Nuzman <joseph.nuzman@intel.com>,
Steven Rostedt <rostedt@goodmis.org>
Subject: Re: [patch 02/38] x86/cpu: Use native_wrmsrl() in load_percpu_segment()
Date: Mon, 18 Jul 2022 07:11:08 +0200 [thread overview]
Message-ID: <70b03d06-6ab9-1693-f811-f784a7dced76@suse.com> (raw)
In-Reply-To: <87o7xnwgl3.ffs@tglx>
[-- Attachment #1.1.1: Type: text/plain, Size: 7163 bytes --]
On 17.07.22 23:54, Thomas Gleixner wrote:
> On Sun, Jul 17 2022 at 22:13, Thomas Gleixner wrote:
>> On Sun, Jul 17 2022 at 22:08, Thomas Gleixner wrote:
>>> On Sun, Jul 17 2022 at 21:08, Thomas Gleixner wrote:
>>> loadsegment_simple() was a red herring. The gs segment is already zero.
>>>
>>> So what explodes here is the early boot when switching from early per
>>> CPU to the real per CPU area.
>>>
>>> start_kernel()
>>> .....
>>> setup_per_cpu_areas();
>>> smp_prepare_boot_cpu()
>>
>> Bah. switch_to_new_gdt() is already invoked from setup_per_cpu_areas()
>> and then again in smp_prepare_boot_cpu() and once more in cpu_init(),
>>
>> What a mess.
>
> So the below builds and boots at least on 64bit. I'll stare at it some
> more tomorrow. I have no idea whether native_load_gdt() works with
> XEN_PV. It should, but what do I know.
No, shouldn't work. But ...
>
> Thanks,
>
> tglx
> ---
> --- a/arch/x86/include/asm/desc.h
> +++ b/arch/x86/include/asm/desc.h
> @@ -205,7 +205,7 @@ static inline void native_set_ldt(const
> }
> }
>
> -static inline void native_load_gdt(const struct desc_ptr *dtr)
> +static __always_inline void native_load_gdt(const struct desc_ptr *dtr)
> {
> asm volatile("lgdt %0"::"m" (*dtr));
> }
> --- a/arch/x86/include/asm/processor.h
> +++ b/arch/x86/include/asm/processor.h
> @@ -670,10 +670,9 @@ extern int sysenter_setup(void);
> /* Defined in head.S */
> extern struct desc_ptr early_gdt_descr;
>
> -extern void switch_to_new_gdt(int);
> +extern void switch_to_real_gdt(int);
> extern void load_direct_gdt(int);
> extern void load_fixmap_gdt(int);
> -extern void load_percpu_segment(int);
> extern void cpu_init(void);
> extern void cpu_init_secondary(void);
> extern void cpu_init_exception_handling(void);
> --- a/arch/x86/kernel/cpu/Makefile
> +++ b/arch/x86/kernel/cpu/Makefile
> @@ -7,20 +7,24 @@
> ifdef CONFIG_FUNCTION_TRACER
> CFLAGS_REMOVE_common.o = -pg
> CFLAGS_REMOVE_perf_event.o = -pg
> +CFLAGS_REMOVE_switch_gdt.o = -pg
> endif
>
> # If these files are instrumented, boot hangs during the first second.
> KCOV_INSTRUMENT_common.o := n
> KCOV_INSTRUMENT_perf_event.o := n
> +KCOV_INSTRUMENT_switch_gdt.o := n
>
> # As above, instrumenting secondary CPU boot code causes boot hangs.
> KCSAN_SANITIZE_common.o := n
> +KCSAN_SANITIZE_switch_gdt.o := n
>
> -# Make sure load_percpu_segment has no stackprotector
> -CFLAGS_common.o := -fno-stack-protector
> +# Make sure that switching the GDT and the per CPU segment
> +# does not have stack protector enabled.
> +CFLAGS_switch_gdt.o := -fno-stack-protector
>
> obj-y := cacheinfo.o scattered.o topology.o
> -obj-y += common.o
> +obj-y += common.o switch_gdt.o
> obj-y += rdrand.o
> obj-y += match.o
> obj-y += bugs.o
> --- a/arch/x86/kernel/cpu/common.c
> +++ b/arch/x86/kernel/cpu/common.c
> @@ -701,16 +701,6 @@ static const char *table_lookup_model(st
> __u32 cpu_caps_cleared[NCAPINTS + NBUGINTS] __aligned(sizeof(unsigned long));
> __u32 cpu_caps_set[NCAPINTS + NBUGINTS] __aligned(sizeof(unsigned long));
>
> -void load_percpu_segment(int cpu)
> -{
> -#ifdef CONFIG_X86_32
> - loadsegment(fs, __KERNEL_PERCPU);
> -#else
> - __loadsegment_simple(gs, 0);
> - wrmsrl(MSR_GS_BASE, cpu_kernelmode_gs_base(cpu));
> -#endif
> -}
> -
> #ifdef CONFIG_X86_32
> /* The 32-bit entry code needs to find cpu_entry_area. */
> DEFINE_PER_CPU(struct cpu_entry_area *, cpu_entry_area);
> @@ -738,18 +728,6 @@ void load_fixmap_gdt(int cpu)
> }
> EXPORT_SYMBOL_GPL(load_fixmap_gdt);
>
> -/*
> - * Current gdt points %fs at the "master" per-cpu area: after this,
> - * it's on the real one.
> - */
> -void switch_to_new_gdt(int cpu)
> -{
> - /* Load the original GDT */
> - load_direct_gdt(cpu);
> - /* Reload the per-cpu base */
> - load_percpu_segment(cpu);
> -}
> -
> static const struct cpu_dev *cpu_devs[X86_VENDOR_NUM] = {};
>
> static void get_model_name(struct cpuinfo_x86 *c)
> @@ -2228,12 +2206,6 @@ void cpu_init(void)
> boot_cpu_has(X86_FEATURE_TSC) || boot_cpu_has(X86_FEATURE_DE))
> cr4_clear_bits(X86_CR4_VME|X86_CR4_PVI|X86_CR4_TSD|X86_CR4_DE);
>
> - /*
> - * Initialize the per-CPU GDT with the boot GDT,
> - * and set up the GDT descriptor:
> - */
> - switch_to_new_gdt(cpu);
> -
> if (IS_ENABLED(CONFIG_X86_64)) {
> loadsegment(fs, 0);
> memset(cur->thread.tls_array, 0, GDT_ENTRY_TLS_ENTRIES * 8);
> --- /dev/null
> +++ b/arch/x86/kernel/cpu/switch_gdt.c
> @@ -0,0 +1,31 @@
> +// SPDX-License-Identifier: GPL-2.0-only
> +#include <asm/processor.h>
> +#include <asm/segment.h>
> +#include <asm/desc.h>
> +
> +/*
> + * Invoked during early boot to switch from early GDT and early per CPU
> + * (%fs on 32bit, GS_BASE on 64bit) to the real GDT and the runtime per CPU
> + * area.
> + *
> + * This has to be done atomic because after switching from early GDT to
> + * the real one any per cpu variable access is going to fault because
> + * %fs resp. GS_BASE is not yet pointing to the real per CPU data.
> + *
> + * As a consequence this uses the native variants of load_gdt() and
> + * wrmsrl(). So XEN_PV has to take the fault and emulate.
> + */
> +void __init switch_to_real_gdt(int cpu)
> +{
> + struct desc_ptr gdt_descr;
> +
> + gdt_descr.address = (long)get_cpu_gdt_rw(cpu);
> + gdt_descr.size = GDT_SIZE - 1;
> + native_load_gdt(&gdt_descr);
> +
> +#ifdef CONFIG_X86_32
> + loadsegment(fs, __KERNEL_PERCPU);
> +#else
> + native_wrmsrl(MSR_GS_BASE, cpu_kernelmode_gs_base(cpu));
> +#endif
> +}
> --- a/arch/x86/kernel/setup_percpu.c
> +++ b/arch/x86/kernel/setup_percpu.c
> @@ -211,7 +211,7 @@ void __init setup_per_cpu_areas(void)
> * area. Reload any changed state for the boot CPU.
> */
> if (!cpu)
> - switch_to_new_gdt(cpu);
> + switch_to_real_gdt(cpu);
> }
>
> /* indicate the early static arrays will soon be gone */
> --- a/arch/x86/kernel/smpboot.c
> +++ b/arch/x86/kernel/smpboot.c
> @@ -1457,7 +1457,11 @@ void arch_thaw_secondary_cpus_end(void)
> void __init native_smp_prepare_boot_cpu(void)
> {
> int me = smp_processor_id();
> - switch_to_new_gdt(me);
> +
> + /* SMP invokes this from setup_per_cpu_areas() */
> + if (!IS_ENABLED(CONFIG_SMP))
> + switch_to_real_gdt(me);
> +
> /* already set me in cpu_online_mask in boot_cpu_init() */
> cpumask_set_cpu(me, cpu_callout_mask);
> cpu_set_state_online(me);
> --- a/arch/x86/xen/enlighten_pv.c
> +++ b/arch/x86/xen/enlighten_pv.c
> @@ -1164,7 +1164,7 @@ static void __init xen_setup_gdt(int cpu
> pv_ops.cpu.write_gdt_entry = xen_write_gdt_entry_boot;
> pv_ops.cpu.load_gdt = xen_load_gdt_boot;
>
> - switch_to_new_gdt(cpu);
> + switch_to_real_gdt(cpu);
... can't you use the paravirt variant of load_gdt in switch_to_real_gdt() ?
>
> pv_ops.cpu.write_gdt_entry = xen_write_gdt_entry;
> pv_ops.cpu.load_gdt = xen_load_gdt;
>
Juergen
[-- Attachment #1.1.2: OpenPGP public key --]
[-- Type: application/pgp-keys, Size: 3149 bytes --]
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 495 bytes --]
next prev parent reply other threads:[~2022-07-18 5:11 UTC|newest]
Thread overview: 142+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-07-16 23:17 [patch 00/38] x86/retbleed: Call depth tracking mitigation Thomas Gleixner
2022-07-16 23:17 ` [patch 01/38] x86/paravirt: Ensure proper alignment Thomas Gleixner
2022-07-16 23:17 ` [patch 02/38] x86/cpu: Use native_wrmsrl() in load_percpu_segment() Thomas Gleixner
2022-07-17 0:22 ` Andrew Cooper
2022-07-17 15:20 ` Linus Torvalds
2022-07-17 19:08 ` Thomas Gleixner
2022-07-17 20:08 ` Thomas Gleixner
2022-07-17 20:13 ` Thomas Gleixner
2022-07-17 21:54 ` Thomas Gleixner
2022-07-18 5:11 ` Juergen Gross [this message]
2022-07-18 6:54 ` Thomas Gleixner
2022-07-18 8:55 ` Thomas Gleixner
2022-07-18 9:31 ` Peter Zijlstra
2022-07-18 10:33 ` Thomas Gleixner
2022-07-18 11:42 ` Thomas Gleixner
2022-07-18 17:52 ` [patch 0/3] x86/cpu: Sanitize switch_to_new_gdt() Thomas Gleixner
2022-07-18 17:52 ` [patch 1/3] x86/cpu: Remove segment load from switch_to_new_gdt() Thomas Gleixner
2022-07-18 18:43 ` Linus Torvalds
2022-07-18 18:55 ` Thomas Gleixner
2022-07-18 17:52 ` [patch 2/3] x86/cpu: Get rid of redundant switch_to_new_gdt() invocations Thomas Gleixner
2022-07-18 17:52 ` [patch 3/3] x86/cpu: Re-enable stackprotector Thomas Gleixner
2022-07-16 23:17 ` [patch 03/38] x86/modules: Set VM_FLUSH_RESET_PERMS in module_alloc() Thomas Gleixner
2022-07-16 23:17 ` [patch 04/38] x86/vdso: Ensure all kernel code is seen by objtool Thomas Gleixner
2022-07-16 23:17 ` [patch 05/38] btree: Initialize early when builtin Thomas Gleixner
2022-07-16 23:17 ` [patch 06/38] objtool: Allow GS relative relocs Thomas Gleixner
2022-07-16 23:17 ` [patch 07/38] objtool: Track init section Thomas Gleixner
2022-07-16 23:17 ` [patch 08/38] objtool: Add .call_sites section Thomas Gleixner
2022-07-16 23:17 ` [patch 09/38] objtool: Add .sym_sites section Thomas Gleixner
2022-07-16 23:17 ` [patch 10/38] objtool: Add --hacks=skylake Thomas Gleixner
2022-07-16 23:17 ` [patch 11/38] objtool: Allow STT_NOTYPE -> STT_FUNC+0 tail-calls Thomas Gleixner
2022-07-16 23:17 ` [patch 12/38] x86/entry: Make sync_regs() invocation a tail call Thomas Gleixner
2022-07-16 23:17 ` [patch 13/38] x86/modules: Make module_alloc() generally available Thomas Gleixner
2022-07-16 23:17 ` [patch 14/38] x86/Kconfig: Add CONFIG_CALL_THUNKS Thomas Gleixner
2022-07-16 23:17 ` [patch 15/38] x86/retbleed: Add X86_FEATURE_CALL_DEPTH Thomas Gleixner
2022-07-16 23:17 ` [patch 16/38] modules: Make struct module_layout unconditionally available Thomas Gleixner
2022-07-16 23:17 ` [patch 17/38] module: Add arch_data to module_layout Thomas Gleixner
2022-07-16 23:17 ` [patch 18/38] mm/vmalloc: Provide huge page mappings Thomas Gleixner
2022-07-16 23:17 ` [patch 19/38] x86/module: Provide __module_alloc() Thomas Gleixner
2022-07-16 23:17 ` [patch 20/38] x86/alternatives: Provide text_poke_[copy|set]_locked() Thomas Gleixner
2022-07-16 23:17 ` [patch 21/38] x86/entry: Make some entry symbols global Thomas Gleixner
2022-07-16 23:17 ` [patch 22/38] x86/paravirt: Make struct paravirt_call_site unconditionally available Thomas Gleixner
2022-07-16 23:17 ` [patch 23/38] x86/callthunks: Add call patching for call depth tracking Thomas Gleixner
2022-07-16 23:17 ` [patch 24/38] module: Add layout for callthunks tracking Thomas Gleixner
2022-07-16 23:17 ` [patch 25/38] x86/modules: Add call thunk patching Thomas Gleixner
2022-07-16 23:17 ` [patch 26/38] x86/returnthunk: Allow different return thunks Thomas Gleixner
2022-07-16 23:17 ` [patch 27/38] x86/asm: Provide ALTERNATIVE_3 Thomas Gleixner
2022-07-16 23:17 ` [patch 28/38] x86/retbleed: Add SKL return thunk Thomas Gleixner
2022-07-16 23:17 ` [patch 29/38] x86/retpoline: Add SKL retthunk retpolines Thomas Gleixner
2022-07-16 23:17 ` [patch 30/38] x86/retbleed: Add SKL call thunk Thomas Gleixner
2022-07-16 23:18 ` [patch 31/38] x86/calldepth: Add ret/call counting for debug Thomas Gleixner
2022-07-16 23:18 ` [patch 32/38] static_call: Add call depth tracking support Thomas Gleixner
2022-07-16 23:18 ` [patch 33/38] kallsyms: Take callthunks into account Thomas Gleixner
2022-07-16 23:18 ` [patch 34/38] x86/orc: Make it callthunk aware Thomas Gleixner
2022-07-16 23:18 ` [patch 35/38] kprobes: Add callthunk blacklisting Thomas Gleixner
2022-07-16 23:18 ` [patch 36/38] x86/ftrace: Make it call depth tracking aware Thomas Gleixner
2022-07-18 21:01 ` Steven Rostedt
2022-07-19 8:46 ` Peter Zijlstra
2022-07-19 13:06 ` Steven Rostedt
2022-07-16 23:18 ` [patch 37/38] x86/bpf: Emit call depth accounting if required Thomas Gleixner
2022-07-19 5:30 ` Alexei Starovoitov
2022-07-19 8:34 ` Peter Zijlstra
2022-07-16 23:18 ` [patch 38/38] x86/retbleed: Add call depth tracking mitigation Thomas Gleixner
2022-07-17 9:45 ` [patch 00/38] x86/retbleed: Call " David Laight
2022-07-17 15:07 ` Thomas Gleixner
2022-07-17 17:56 ` David Laight
2022-07-17 19:15 ` Thomas Gleixner
2022-07-18 19:29 ` Thomas Gleixner
2022-07-18 19:30 ` Thomas Gleixner
2022-07-18 19:51 ` Linus Torvalds
2022-07-18 20:44 ` Thomas Gleixner
2022-07-18 21:01 ` Linus Torvalds
2022-07-18 21:43 ` Peter Zijlstra
2022-07-18 22:34 ` Linus Torvalds
2022-07-18 23:52 ` Peter Zijlstra
2022-07-18 21:18 ` Peter Zijlstra
2022-07-18 22:22 ` Thomas Gleixner
2022-07-18 22:47 ` Joao Moreira
2022-07-18 22:55 ` Sami Tolvanen
2022-07-18 23:08 ` Joao Moreira
2022-07-18 23:19 ` Thomas Gleixner
2022-07-18 23:42 ` Linus Torvalds
2022-07-18 23:52 ` Linus Torvalds
2022-07-18 23:57 ` Peter Zijlstra
2022-07-19 0:03 ` Linus Torvalds
2022-07-19 0:11 ` Linus Torvalds
2022-07-19 0:23 ` Peter Zijlstra
2022-07-19 1:02 ` Linus Torvalds
2022-07-19 17:19 ` Sami Tolvanen
2022-07-20 21:13 ` Peter Zijlstra
2022-07-21 8:21 ` David Laight
2022-07-21 10:56 ` David Laight
2022-07-21 15:54 ` Peter Zijlstra
2022-07-21 17:55 ` Peter Zijlstra
2022-07-21 18:06 ` Linus Torvalds
2022-07-21 18:27 ` Peter Zijlstra
2022-07-21 18:32 ` Linus Torvalds
2022-07-21 20:22 ` Joao Moreira
2022-07-22 0:16 ` Sami Tolvanen
2022-07-22 10:23 ` Peter Zijlstra
2022-07-22 15:38 ` Sami Tolvanen
2022-07-21 22:01 ` David Laight
2022-07-22 11:03 ` Peter Zijlstra
2022-07-22 13:27 ` David Laight
2022-07-23 9:50 ` Thomas Gleixner
2022-07-19 0:01 ` Linus Torvalds
2022-07-19 0:19 ` Joao Moreira
2022-07-19 17:21 ` Sami Tolvanen
2022-07-19 17:58 ` Joao Moreira
2022-07-19 8:26 ` David Laight
2022-07-19 16:27 ` Linus Torvalds
2022-07-19 17:23 ` Sami Tolvanen
2022-07-19 17:27 ` Linus Torvalds
2022-07-19 18:06 ` Sami Tolvanen
2022-07-19 20:10 ` Peter Zijlstra
2022-07-18 22:48 ` Sami Tolvanen
2022-07-18 22:59 ` Thomas Gleixner
2022-07-18 23:10 ` Sami Tolvanen
2022-07-18 23:39 ` Linus Torvalds
2022-07-18 23:51 ` Peter Zijlstra
2022-07-20 9:00 ` Thomas Gleixner
2022-07-20 16:55 ` Sami Tolvanen
2022-07-20 19:42 ` Sami Tolvanen
2022-07-22 20:11 ` Tim Chen
2022-07-22 22:18 ` Linus Torvalds
2022-07-18 19:55 ` Thomas Gleixner
2022-07-19 10:24 ` Virt " Andrew Cooper
2022-07-19 14:13 ` Thomas Gleixner
2022-07-19 16:23 ` Andrew Cooper
2022-07-19 21:17 ` Thomas Gleixner
2022-07-19 14:45 ` Michael Kelley (LINUX)
2022-07-19 20:16 ` Peter Zijlstra
2022-07-20 16:57 ` [patch 00/38] x86/retbleed: " Steven Rostedt
2022-07-20 17:09 ` Linus Torvalds
2022-07-20 17:24 ` Peter Zijlstra
2022-07-20 17:50 ` Steven Rostedt
2022-07-20 18:07 ` Linus Torvalds
2022-07-20 18:31 ` Steven Rostedt
2022-07-20 18:43 ` Linus Torvalds
2022-07-20 19:11 ` Steven Rostedt
2022-07-20 19:36 ` Kees Cook
2022-07-20 19:43 ` Steven Rostedt
2022-07-20 21:36 ` Peter Zijlstra
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=70b03d06-6ab9-1693-f811-f784a7dced76@suse.com \
--to=jgross@suse.com \
--cc=Andrew.Cooper3@citrix.com \
--cc=alyssa.milburn@linux.intel.com \
--cc=hjl.tools@gmail.com \
--cc=jannh@google.com \
--cc=joao.moreira@intel.com \
--cc=joseph.nuzman@intel.com \
--cc=jpoimboe@kernel.org \
--cc=kwikner@ethz.ch \
--cc=linux-kernel@vger.kernel.org \
--cc=pawan.kumar.gupta@linux.intel.com \
--cc=rostedt@goodmis.org \
--cc=tglx@linutronix.de \
--cc=tim.c.chen@linux.intel.com \
--cc=torvalds@linux-foundation.org \
--cc=x86@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).