linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Andy Lutomirski <luto@kernel.org>
To: x86@kernel.org, linux-kernel@vger.kernel.org
Cc: linux-arch@vger.kernel.org, Borislav Petkov <bp@alien8.de>,
	Nadav Amit <nadav.amit@gmail.com>,
	Kees Cook <keescook@chromium.org>,
	Brian Gerst <brgerst@gmail.com>,
	"kernel-hardening@lists.openwall.com" 
	<kernel-hardening@lists.openwall.com>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Josh Poimboeuf <jpoimboe@redhat.com>, Jann Horn <jann@thejh.net>,
	Heiko Carstens <heiko.carstens@de.ibm.com>,
	Andy Lutomirski <luto@kernel.org>
Subject: [PATCH v5 32/32] fork: Cache two thread stacks per cpu if CONFIG_VMAP_STACK is set
Date: Mon, 11 Jul 2016 13:54:05 -0700	[thread overview]
Message-ID: <ba60f1faab6c4469853f5bc31d1852c455268763.1468270393.git.luto@kernel.org> (raw)
In-Reply-To: <cover.1468270393.git.luto@kernel.org>
In-Reply-To: <cover.1468270393.git.luto@kernel.org>

vmalloc is a bit slow, and pounding vmalloc/vfree will eventually
force a global TLB flush.

To reduce pressure on them, if CONFIG_VMAP_STACK, cache two thread
stacks per cpu.  This will let us quickly allocate a hopefully
cache-hot, TLB-hot stack under heavy forking workloads (shell script
style).

On my silly pthread_create benchmark, it saves about 2 µs per
pthread_create+join with CONFIG_VMAP_STACK=y.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
---
 kernel/fork.c | 62 ++++++++++++++++++++++++++++++++++++++++++++++++++---------
 1 file changed, 53 insertions(+), 9 deletions(-)

diff --git a/kernel/fork.c b/kernel/fork.c
index 3003edae90fc..e8e21134ba23 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -159,15 +159,41 @@ void __weak arch_release_thread_stack(unsigned long *stack)
  * kmemcache based allocator.
  */
 # if THREAD_SIZE >= PAGE_SIZE || defined(CONFIG_VMAP_STACK)
+
+#ifdef CONFIG_VMAP_STACK
+/*
+ * vmalloc is a bit slow, and calling vfree enough times will force a TLB
+ * flush.  Try to minimize the number of calls by caching stacks.
+ */
+#define NR_CACHED_STACKS 2
+static DEFINE_PER_CPU(struct vm_struct *, cached_stacks[NR_CACHED_STACKS]);
+#endif
+
 static unsigned long *alloc_thread_stack_node(struct task_struct *tsk, int node)
 {
 #ifdef CONFIG_VMAP_STACK
-	void *stack = __vmalloc_node_range(THREAD_SIZE, THREAD_SIZE,
-					   VMALLOC_START, VMALLOC_END,
-					   THREADINFO_GFP | __GFP_HIGHMEM,
-					   PAGE_KERNEL,
-					   0, node,
-					   __builtin_return_address(0));
+	void *stack;
+	int i;
+
+	local_irq_disable();
+	for (i = 0; i < NR_CACHED_STACKS; i++) {
+		struct vm_struct *s = this_cpu_read(cached_stacks[i]);
+
+		if (!s)
+			continue;
+		this_cpu_write(cached_stacks[i], NULL);
+
+		tsk->stack_vm_area = s;
+		local_irq_enable();
+		return s->addr;
+	}
+	local_irq_enable();
+
+	stack = __vmalloc_node_range(THREAD_SIZE, THREAD_SIZE,
+				     VMALLOC_START, VMALLOC_END,
+				     THREADINFO_GFP | __GFP_HIGHMEM,
+				     PAGE_KERNEL,
+				     0, node, __builtin_return_address(0));
 
 	/*
 	 * We can't call find_vm_area() in interrupt context, and
@@ -187,10 +213,28 @@ static unsigned long *alloc_thread_stack_node(struct task_struct *tsk, int node)
 
 static inline void free_thread_stack(struct task_struct *tsk)
 {
-	if (task_stack_vm_area(tsk))
+#ifdef CONFIG_VMAP_STACK
+	if (task_stack_vm_area(tsk)) {
+		unsigned long flags;
+		int i;
+
+		local_irq_save(flags);
+		for (i = 0; i < NR_CACHED_STACKS; i++) {
+			if (this_cpu_read(cached_stacks[i]))
+				continue;
+
+			this_cpu_write(cached_stacks[i], tsk->stack_vm_area);
+			local_irq_restore(flags);
+			return;
+		}
+		local_irq_restore(flags);
+
 		vfree(tsk->stack);
-	else
-		free_kmem_pages((unsigned long)tsk->stack, THREAD_SIZE_ORDER);
+		return;
+	}
+#endif
+
+	free_kmem_pages((unsigned long)tsk->stack, THREAD_SIZE_ORDER);
 }
 # else
 static struct kmem_cache *thread_stack_cache;
-- 
2.7.4

  parent reply	other threads:[~2016-07-11 20:55 UTC|newest]

Thread overview: 58+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-07-11 20:53 [PATCH v5 00/32] virtually mapped stacks and thread_info cleanup Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 01/32] bluetooth: Switch SMP to crypto_cipher_encrypt_one() Andy Lutomirski
2016-07-14 19:10   ` Andy Lutomirski
2016-07-14 20:30     ` Marcel Holtmann
2016-07-14 20:41     ` David Miller
2016-07-11 20:53 ` [PATCH v5 02/32] x86/mm/hotplug: Don't remove PGD entries in remove_pagetable() Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 03/32] x86/cpa: In populate_pgd, don't set the pgd entry until it's populated Andy Lutomirski
2016-07-22  4:43   ` [kernel-hardening] " Valdis.Kletnieks
2016-07-22  5:34     ` Andy Lutomirski
2016-07-22 10:21       ` Ingo Molnar
2016-07-22 18:21         ` Andy Lutomirski
2016-07-22 18:31           ` Andy Lutomirski
2016-07-22 20:11           ` Ingo Molnar
2016-07-23  5:21       ` Valdis.Kletnieks
2016-07-23 14:58         ` Nicolai Stange
2016-07-28  9:26           ` Valdis.Kletnieks
2016-07-11 20:53 ` [PATCH v5 04/32] x86/mm: Remove kernel_unmap_pages_in_pgd() and efi_cleanup_page_tables() Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 05/32] mm: Track NR_KERNEL_STACK in KiB instead of number of stacks Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 06/32] mm: Fix memcg stack accounting for sub-page stacks Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 07/32] fork: Add generic vmalloced stack support Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 08/32] dma-api: Teach the "DMA-from-stack" check about vmapped stacks Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 09/32] x86/dumpstack: When OOPSing, rewind the stack before do_exit() Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 10/32] x86/dumpstack: Honor supplied @regs arg Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 11/32] x86/dumpstack: Try harder to get a call trace on stack overflow Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 12/32] x86/dumpstack/64: Handle faults when printing the "Stack:" part of an OOPS Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 13/32] x86/mm/64: In vmalloc_fault(), use CR3 instead of current->active_mm Andy Lutomirski
2016-07-12 17:51   ` [kernel-hardening] " Dave Hansen
2016-07-12 18:03     ` Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 14/32] x86/mm/64: Enable vmapped stacks Andy Lutomirski
2016-07-13  7:53   ` Ingo Molnar
2016-07-13 18:42     ` Andy Lutomirski
2016-07-14  8:34       ` Ingo Molnar
2016-07-14 16:51         ` Andy Lutomirski
2016-07-14 18:45           ` Ingo Molnar
2016-07-11 20:53 ` [PATCH v5 15/32] x86/mm: Improve stack-overflow #PF handling Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 16/32] x86: Move uaccess_err and sig_on_uaccess_err to thread_struct Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 17/32] x86: Move addr_limit " Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 18/32] signal: Consolidate {TS,TLF}_RESTORE_SIGMASK code Andy Lutomirski
2016-07-12 11:57   ` Brian Gerst
2016-07-12 23:01     ` Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 19/32] x86/smp: Remove stack_smp_processor_id() Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 20/32] x86/smp: Remove unnecessary initialization of thread_info::cpu Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 21/32] x86/asm: Move 'status' from struct thread_info to struct thread_struct Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 22/32] kdb: Use task_cpu() instead of task_thread_info()->cpu Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 23/32] printk: When dumping regs, show the stack, not thread_info Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 24/32] x86/entry: Get rid of pt_regs_to_thread_info() Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 25/32] um: Stop conflating task_struct::stack with thread_info Andy Lutomirski
2016-07-11 20:53 ` [PATCH v5 26/32] sched: Allow putting thread_info into task_struct Andy Lutomirski
2016-07-11 20:54 ` [PATCH v5 27/32] x86: Move " Andy Lutomirski
2016-07-11 20:54 ` [PATCH v5 28/32] sched: Add try_get_task_stack() and put_task_stack() Andy Lutomirski
2016-07-11 20:54 ` [PATCH v5 29/32] kthread: to_live_kthread() needs try_get_task_stack() Andy Lutomirski
2016-07-11 20:54 ` [PATCH v5 30/32] x86/dumpstack: Pin the target stack in save_stack_trace_tsk() Andy Lutomirski
2016-07-11 20:54 ` [PATCH v5 31/32] sched: Free the stack early if CONFIG_THREAD_INFO_IN_TASK Andy Lutomirski
2016-07-11 20:54 ` Andy Lutomirski [this message]
2016-07-12  8:56 ` [PATCH v5 00/32] virtually mapped stacks and thread_info cleanup Herbert Xu
2016-07-13  8:54 ` Christian Borntraeger
2016-07-13 18:36   ` Andy Lutomirski
2016-07-13 18:53     ` Christian Borntraeger

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=ba60f1faab6c4469853f5bc31d1852c455268763.1468270393.git.luto@kernel.org \
    --to=luto@kernel.org \
    --cc=bp@alien8.de \
    --cc=brgerst@gmail.com \
    --cc=heiko.carstens@de.ibm.com \
    --cc=jann@thejh.net \
    --cc=jpoimboe@redhat.com \
    --cc=keescook@chromium.org \
    --cc=kernel-hardening@lists.openwall.com \
    --cc=linux-arch@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=nadav.amit@gmail.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).