From: "Paul E. McKenney" <paulmck@kernel.org>
To: Thomas Gleixner <tglx@linutronix.de>
Cc: LKML <linux-kernel@vger.kernel.org>,
x86@kernel.org, Andy Lutomirski <luto@kernel.org>,
Alexandre Chartre <alexandre.chartre@oracle.com>,
Frederic Weisbecker <frederic@kernel.org>,
Paolo Bonzini <pbonzini@redhat.com>,
Sean Christopherson <sean.j.christopherson@intel.com>,
Masami Hiramatsu <mhiramat@kernel.org>,
Petr Mladek <pmladek@suse.com>,
Steven Rostedt <rostedt@goodmis.org>,
Joel Fernandes <joel@joelfernandes.org>,
Boris Ostrovsky <boris.ostrovsky@oracle.com>,
Juergen Gross <jgross@suse.com>, Brian Gerst <brgerst@gmail.com>,
Mathieu Desnoyers <mathieu.desnoyers@efficios.com>,
Josh Poimboeuf <jpoimboe@redhat.com>,
Will Deacon <will@kernel.org>,
Tom Lendacky <thomas.lendacky@amd.com>,
Wei Liu <wei.liu@kernel.org>,
Michael Kelley <mikelley@microsoft.com>,
Jason Chen CJ <jason.cj.chen@intel.com>,
Zhao Yakui <yakui.zhao@intel.com>,
"Peter Zijlstra (Intel)" <peterz@infradead.org>
Subject: Re: [patch V5 00/38] x86/entry: Entry/exception code rework - the leftovers
Date: Thu, 14 May 2020 09:35:20 -0700 [thread overview]
Message-ID: <20200514163520.GA15870@paulmck-ThinkPad-P72> (raw)
In-Reply-To: <20200512210059.056244513@linutronix.de>
On Tue, May 12, 2020 at 11:00:59PM +0200, Thomas Gleixner wrote:
> Folks!
>
> This is V5 of the rework series. V4 can be found here:
>
> https://lore.kernel.org/r/20200505135341.730586321@linutronix.de
>
> Some of the non-x86 specific parts from part 1- 4 have been applied to the
> relevant tip branches and I merged them all together to base the rest on.
>
> I have applied most of the patches from part 1-4 on top after fixing up the
> small review comments all over the place. The result can be found here:
>
> git://git.kernel.org/pub/scm/linux/kernel/git/tglx/devel.git entry-base-v5
>
> I dropped:
>
> [patch V4 part 1 33/36] x86,tracing: Robustify ftrace_nmi_enter()
> [patch V4 part 2 10/18] x86/entry/64: Check IF in __preempt_enable_notrace()
> [patch V4 part 2 11/18] x86/entry/64: Mark ___preempt_schedule_notrace()
>
> from V4 because Peter's attempt to sanitize ftrace_nmi_enter/exit() and the
> underlying callchains turned out to be incomplete and cause the whole
> __preempt_schedule_notrace() hackery which is caused by a gazillion of
> preempt_disable_notrace() / preempt_enable_notrace() invocations in the
> full call chain. I found a better way to do that and that's in part 5 now.
>
> The remaining patches are mostly from part 5 of version 4 dealing with the
> #PF, device interrupts and the system vectors.
>
> The main differences to V4 part 5 are:
>
> - Provide nmi_enter/exit_notrace() which do not contain the
> ftrace_nmi_enter/exit() calls, use them in x86 and make the ftrace
> invocation explicit at a safe place.
>
> - A fix for the reworked SVM code
>
> - An optimization of asm_native_load_gs_index() which I found while
> staring at stuff in more detail
>
> - The removal of the x86 specific irq_regs implementation which was
> aside of two underscores in the variable name exactly the same thing
> as the generic variant.
>
> - Overhaul of the IRQ stack switching mechanics to avoid the objtool and
> unwinder nastiness of doing it in C inline ASM. As discussed here:
>
> https://lore.kernel.org/r/20200507161020.783541450@infradead.org
>
> We agreed on having the actual stack switch still in ASM and utilize it
> for device interrupts, soft interrupts and system vectors.
>
> This comes for the price of an indirect call, but the device interrupt
> handling has an indirect call and soft interrupts and most system
> vectors are indirect call laden already.
>
> There is a lightweight variant for the system vectors which do
> basically nothing, i.e. reschedule IPI and the KVM posted interrupt
> vectors. That variant runs on the interrupted stack and avoids the
> full entry/exit handling overhead by utilizing the conditional RCU
> variant of idtentry_enter/exit().
>
> - Addressing a few review comments.
>
> The series is also available from git:
>
> git://git.kernel.org/pub/scm/linux/kernel/git/tglx/devel.git entry-v5-the-rest
>
> Running the new objtool checks results in two warning types:
>
> - The MCE code has an issue in the non-instrumentable section which is
> still under investigation and discussion
>
> - When paravirt is enabled then there are a few complaints about
> invoking indirect calls (PVOPS) which leave the noinstr sections
> but they are halfways straight forward to fix.
>
> Thanks,
>
> tglx
FWIW, the full set of one-hour rcutorture scenarios completed without
complaint on entry-v5-the-rest. Must be some mistake. ;-)
This is of course all within qemu/KVM guest OSes, so the next step
is to try it on bare metal.
Thanx, Paul
> 8<--------------
>
> arch/x86/include/asm/acrn.h | 11
> arch/x86/include/asm/entry_arch.h | 56 --
> arch/x86/include/asm/irq_regs.h | 32 -
> b/arch/x86/entry/calling.h | 25 -
> b/arch/x86/entry/common.c | 216 +++++++++-
> b/arch/x86/entry/entry_32.S | 260 +-----------
> b/arch/x86/entry/entry_64.S | 581 +++--------------------------
> b/arch/x86/entry/thunk_64.S | 9
> b/arch/x86/hyperv/hv_init.c | 9
> b/arch/x86/include/asm/apic.h | 33 -
> b/arch/x86/include/asm/hw_irq.h | 22 -
> b/arch/x86/include/asm/idtentry.h | 262 ++++++++++++-
> b/arch/x86/include/asm/irq.h | 6
> b/arch/x86/include/asm/irq_stack.h | 40 +
> b/arch/x86/include/asm/irq_work.h | 1
> b/arch/x86/include/asm/irqflags.h | 10
> b/arch/x86/include/asm/mshyperv.h | 13
> b/arch/x86/include/asm/trace/common.h | 4
> b/arch/x86/include/asm/trace/irq_vectors.h | 17
> b/arch/x86/include/asm/traps.h | 21 -
> b/arch/x86/include/asm/uv/uv_bau.h | 8
> b/arch/x86/kernel/apic/apic.c | 39 -
> b/arch/x86/kernel/apic/msi.c | 3
> b/arch/x86/kernel/apic/vector.c | 5
> b/arch/x86/kernel/cpu/acrn.c | 9
> b/arch/x86/kernel/cpu/mce/amd.c | 5
> b/arch/x86/kernel/cpu/mce/core.c | 4
> b/arch/x86/kernel/cpu/mce/therm_throt.c | 5
> b/arch/x86/kernel/cpu/mce/threshold.c | 5
> b/arch/x86/kernel/cpu/mshyperv.c | 22 -
> b/arch/x86/kernel/head_64.S | 7
> b/arch/x86/kernel/idt.c | 42 +-
> b/arch/x86/kernel/irq.c | 58 --
> b/arch/x86/kernel/irq_64.c | 17
> b/arch/x86/kernel/irq_work.c | 6
> b/arch/x86/kernel/kvm.c | 14
> b/arch/x86/kernel/nmi.c | 9
> b/arch/x86/kernel/smp.c | 37 -
> b/arch/x86/kernel/tracepoint.c | 17
> b/arch/x86/kernel/traps.c | 10
> b/arch/x86/kvm/svm/svm.c | 2
> b/arch/x86/mm/fault.c | 69 ++-
> b/arch/x86/platform/uv/tlb_uv.c | 2
> b/arch/x86/xen/enlighten_hvm.c | 12
> b/arch/x86/xen/enlighten_pv.c | 2
> b/arch/x86/xen/setup.c | 4
> b/arch/x86/xen/smp_pv.c | 3
> b/arch/x86/xen/xen-asm_32.S | 12
> b/arch/x86/xen/xen-asm_64.S | 4
> b/arch/x86/xen/xen-ops.h | 1
> b/drivers/xen/events/events_base.c | 6
> b/drivers/xen/preempt.c | 2
> b/include/linux/hardirq.h | 51 ++
> b/include/linux/rcutiny.h | 1
> b/include/linux/rcutree.h | 1
> b/include/xen/events.h | 7
> b/include/xen/xen-ops.h | 7
> b/kernel/rcu/tree.c | 5
> b/kernel/softirq.c | 35 +
> 59 files changed, 906 insertions(+), 1270 deletions(-)
>
prev parent reply other threads:[~2020-05-14 16:35 UTC|newest]
Thread overview: 55+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-05-12 21:00 [patch V5 00/38] x86/entry: Entry/exception code rework - the leftovers Thomas Gleixner
2020-05-12 21:01 ` [patch V5 01/38] x86/kvm/svm: Use uninstrumented wrmsrl() to restore GS Thomas Gleixner
2020-05-13 7:11 ` Jürgen Groß
2020-05-12 21:01 ` [patch V5 02/38] x86/entry/64: Use native swapgs in asm_native_load_gs_index() Thomas Gleixner
2020-05-13 2:02 ` Steven Rostedt
2020-05-13 6:34 ` Thomas Gleixner
2020-05-13 7:12 ` Jürgen Groß
2020-05-19 19:58 ` [tip: x86/entry] x86/entry/64: Use native swapgs in asm_load_gs_index() tip-bot2 for Thomas Gleixner
2020-05-12 21:01 ` [patch V5 03/38] nmi, tracing: Provide nmi_enter/exit_notrace() Thomas Gleixner
2020-05-15 1:32 ` Steven Rostedt
2020-05-15 1:35 ` Steven Rostedt
2020-05-15 1:37 ` Steven Rostedt
2020-05-12 21:01 ` [patch V5 04/38] x86: Make hardware latency tracing explicit Thomas Gleixner
2020-05-15 1:43 ` Steven Rostedt
2020-05-15 15:08 ` Thomas Gleixner
2020-05-12 21:01 ` [patch V5 05/38] genirq: Provide irq_enter/exit_rcu() Thomas Gleixner
2020-05-12 21:01 ` [patch V5 06/38] x86/entry: Provide helpers for execute on irqstack Thomas Gleixner
2020-05-13 21:43 ` Josh Poimboeuf
2020-05-12 21:01 ` [patch V5 07/38] x86/entry/64: Move do_softirq_own_stack() to C Thomas Gleixner
2020-05-12 21:01 ` [patch V5 08/38] x86/entry: Split idtentry_enter/exit() Thomas Gleixner
2020-05-12 21:01 ` [patch V5 09/38] x86/entry: Switch XEN/PV hypercall entry to IDTENTRY Thomas Gleixner
2020-05-14 16:24 ` Boris Ostrovsky
2020-05-12 21:01 ` [patch V5 10/38] x86/entry/64: Simplify idtentry_body Thomas Gleixner
2020-05-12 21:01 ` [patch V5 11/38] rcu: Provide __rcu_is_watching() Thomas Gleixner
2020-05-19 19:52 ` [tip: core/rcu] " tip-bot2 for Thomas Gleixner
2020-05-12 21:01 ` [patch V5 12/38] x86/entry: Provide idtentry_entry/exit_cond_rcu() Thomas Gleixner
2020-05-12 21:01 ` [patch V5 13/38] x86/entry: Switch page fault exception to IDTENTRY_RAW Thomas Gleixner
2020-05-12 21:01 ` [patch V5 14/38] x86/entry: Remove the transition leftovers Thomas Gleixner
2020-05-12 21:01 ` [patch V5 15/38] x86/entry: Change exit path of xen_failsafe_callback Thomas Gleixner
2020-05-12 21:01 ` [patch V5 16/38] x86/entry/64: Remove error_exit Thomas Gleixner
2020-05-12 21:01 ` [patch V5 17/38] x86/entry/32: Remove common_exception Thomas Gleixner
2020-05-12 21:01 ` [patch V5 18/38] x86/irq: Use generic irq_regs implementation Thomas Gleixner
2020-05-12 21:01 ` [patch V5 19/38] x86/irq: Convey vector as argument and not in ptregs Thomas Gleixner
2020-05-12 21:01 ` [patch V5 20/38] x86/irq/64: Provide handle_irq() Thomas Gleixner
2020-05-12 21:01 ` [patch V5 21/38] x86/entry: Add IRQENTRY_IRQ macro Thomas Gleixner
2020-05-12 21:01 ` [patch V5 22/38] x86/entry: Use idtentry for interrupts Thomas Gleixner
2020-05-12 21:01 ` [patch V5 23/38] genirq: Provde __irq_enter/exit_raw() Thomas Gleixner
2020-05-12 21:01 ` [patch V5 24/38] x86/entry: Provide IDTENTRY_SYSVEC Thomas Gleixner
2020-05-15 0:16 ` Boris Ostrovsky
2020-05-15 8:52 ` Thomas Gleixner
2020-05-12 21:01 ` [patch V5 25/38] x86/entry: Convert APIC interrupts to IDTENTRY_SYSVEC Thomas Gleixner
2020-05-12 21:01 ` [patch V5 26/38] x86/entry: Convert SMP system vectors " Thomas Gleixner
2020-05-12 21:01 ` [patch V5 27/38] x86/entry: Convert various system vectors Thomas Gleixner
2020-05-12 21:01 ` [patch V5 28/38] x86/entry: Convert KVM vectors to IDTENTRY_SYSVEC Thomas Gleixner
2020-05-12 21:01 ` [patch V5 29/38] x86/entry: Convert various hypervisor " Thomas Gleixner
2020-05-12 21:01 ` [patch V5 30/38] x86/entry: Convert XEN hypercall vector " Thomas Gleixner
2020-05-12 21:01 ` [patch V5 31/38] x86/entry: Convert reschedule interrupt to IDTENTRY_SYSVEC_SIMPLE Thomas Gleixner
2020-05-12 21:01 ` [patch V5 32/38] x86/entry: Remove the apic/BUILD interrupt leftovers Thomas Gleixner
2020-05-12 21:01 ` [patch V5 33/38] x86/entry/64: Remove IRQ stack switching ASM Thomas Gleixner
2020-05-12 21:01 ` [patch V5 34/38] x86/entry: Make enter_from_user_mode() static Thomas Gleixner
2020-05-12 21:01 ` [patch V5 35/38] x86/entry/32: Remove redundant irq disable code Thomas Gleixner
2020-05-12 21:01 ` [patch V5 36/38] x86/entry/64: Remove TRACE_IRQS_*_DEBUG Thomas Gleixner
2020-05-12 21:01 ` [patch V5 37/38] x86/entry: Move paranoid irq tracing out of ASM code Thomas Gleixner
2020-05-12 21:01 ` [patch V5 38/38] x86/entry: Remove the TRACE_IRQS cruft Thomas Gleixner
2020-05-14 16:35 ` Paul E. McKenney [this message]
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=20200514163520.GA15870@paulmck-ThinkPad-P72 \
--to=paulmck@kernel.org \
--cc=alexandre.chartre@oracle.com \
--cc=boris.ostrovsky@oracle.com \
--cc=brgerst@gmail.com \
--cc=frederic@kernel.org \
--cc=jason.cj.chen@intel.com \
--cc=jgross@suse.com \
--cc=joel@joelfernandes.org \
--cc=jpoimboe@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=luto@kernel.org \
--cc=mathieu.desnoyers@efficios.com \
--cc=mhiramat@kernel.org \
--cc=mikelley@microsoft.com \
--cc=pbonzini@redhat.com \
--cc=peterz@infradead.org \
--cc=pmladek@suse.com \
--cc=rostedt@goodmis.org \
--cc=sean.j.christopherson@intel.com \
--cc=tglx@linutronix.de \
--cc=thomas.lendacky@amd.com \
--cc=wei.liu@kernel.org \
--cc=will@kernel.org \
--cc=x86@kernel.org \
--cc=yakui.zhao@intel.com \
/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).