Messages in this thread | | | Subject | Re: [GIT pull] x86/pti: The real thing | From | Mike Galbraith <> | Date | Sat, 30 Dec 2017 10:04:51 +0100 |
| |
On Fri, 2017-12-29 at 17:42 -0800, Linus Torvalds wrote: > On Thu, Dec 28, 2017 at 12:34 PM, Thomas Gleixner <tglx@linutronix.de> wrote: > > > > please pull the latest x86-pti-for-linus git tree from: > > > > git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git x86-pti-for-linus > > > > This is the final set of enabling page table isolation on x86: > > Ok, after that late MCORE2 scare that held things up, this is now > merged in my tree and pushed out. "WorksForMe(tm)".
But it definitely ain't free. NFS hanging out in the kernel for 82ms at a whack may not be such a bad idea after all ;-)
i4790+tbench
CONFIG_PAGE_TABLE_ISOLATION=y Throughput 3467.72 MB/sec 8 clients 8 procs max_latency=3.499 ms
CONFIG_PAGE_TABLE_ISOLATION=n Throughput 3700.81 MB/sec 8 clients 8 procs max_latency=4.893 ms
CONFIG_PAGE_TABLE_ISOLATION=y CONFIG_PAGE_TABLE_ISOLATION=n 4.48% [k] syscall_return_via_sysret 3.56% [k] copy_user_enhanced_fast_string 3.41% [k] copy_user_enhanced_fast_string 2.54% [.] child_run 2.48% [.] child_run 1.76% [k] tcp_transmit_skb 1.64% [.] __strchr_sse2 1.74% [k] tcp_recvmsg 1.60% [k] tcp_transmit_skb 1.69% [.] __strchr_sse2 1.55% [k] tcp_recvmsg 1.56% [.] __GI_____strtoll_l_internal 1.46% [k] ipt_do_table 1.44% [k] ipt_do_table 1.45% [.] __GI_____strtoll_l_internal 1.42% [k] tcp_sendmsg_locked 1.25% [k] tcp_in_window 1.34% [k] tcp_write_xmit 1.21% [.] next_token 1.34% [k] tcp_clean_rtx_queue 1.17% [k] tcp_clean_rtx_queue 1.26% [.] next_token 1.16% [k] tcp_sendmsg_locked 1.24% [k] tcp_in_window 1.15% [k] __switch_to 1.19% [k] nf_conntrack_in 1.10% [k] switch_mm_irqs_off 1.17% [k] __switch_to 1.09% [k] tcp_write_xmit 1.13% [.] __strcasecmp_l_avx 1.09% [k] tcp_ack 1.10% [k] switch_mm_irqs_off 1.08% [k] nf_hook_slow 1.08% [k] tcp_v4_rcv 1.04% [k] __sched_text_start 1.06% [k] nf_hook_slow 0.99% [.] __strcasecmp_l_avx 1.05% [k] tcp_ack
| |