svn commit: r237433 - in head/sys: amd64/include arm/include conf i386/include ia64/include kern mips/include pc98/include powerpc/include sparc64/include sys x86/include x86/x86
Author: kib Date: Fri Jun 22 07:06:40 2012 New Revision: 237433 URL: http://svn.freebsd.org/changeset/base/237433 Log: Implement mechanism to export some kernel timekeeping data to usermode, using shared page. The structures and functions have vdso prefix, to indicate the intended location of the code in some future. The versioned per-algorithm data is exported in the format of struct vdso_timehands, which mostly repeats the content of in-kernel struct timehands. Usermode reading of the structure can be lockless. Compatibility export for 32bit processes on 64bit host is also provided. Kernel also provides usermode with indication about currently used timecounter, so that libc can fall back to syscall if configured timecounter is unknown to usermode code. The shared data updates are initiated both from the tc_windup(), where a fast task is queued to do the update, and from sysctl handlers which change timecounter. A manual override switch kern.timecounter.fast_gettime allows to turn off the mechanism. Only x86 architectures export the real algorithm data, and there, only for tsc timecounter. HPET counters page could be exported as well, but I prefer to not further glue the kernel and libc ABI there until proper vdso-based solution is developed. Minimal stubs neccessary for non-x86 architectures to still compile are provided. Discussed with: bde Reviewed by: jhb Tested by:flo MFC after:1 month Added: head/sys/amd64/include/vdso.h (contents, props changed) head/sys/arm/include/vdso.h (contents, props changed) head/sys/i386/include/vdso.h (contents, props changed) head/sys/ia64/include/vdso.h (contents, props changed) head/sys/kern/subr_dummy_vdso_tc.c (contents, props changed) head/sys/mips/include/vdso.h (contents, props changed) head/sys/pc98/include/vdso.h (contents, props changed) head/sys/powerpc/include/vdso.h (contents, props changed) head/sys/sparc64/include/vdso.h (contents, props changed) head/sys/sys/vdso.h (contents, props changed) head/sys/x86/include/vdso.h (contents, props changed) Modified: head/sys/conf/files.arm head/sys/conf/files.ia64 head/sys/conf/files.mips head/sys/conf/files.powerpc head/sys/conf/files.sparc64 head/sys/kern/imgact_elf.c head/sys/kern/kern_exec.c head/sys/kern/kern_tc.c head/sys/sys/sysent.h head/sys/x86/x86/tsc.c Added: head/sys/amd64/include/vdso.h == --- /dev/null 00:00:00 1970 (empty, because file is newly added) +++ head/sys/amd64/include/vdso.h Fri Jun 22 07:06:40 2012 (r237433) @@ -0,0 +1,6 @@ +/*- + * This file is in the public domain. + */ +/* $FreeBSD$ */ + +#include Added: head/sys/arm/include/vdso.h == --- /dev/null 00:00:00 1970 (empty, because file is newly added) +++ head/sys/arm/include/vdso.h Fri Jun 22 07:06:40 2012(r237433) @@ -0,0 +1,34 @@ +/*- + * Copyright 2012 Konstantin Belousov . + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + *notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + *notice, this list of conditions and the following disclaimer in the + *documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + * + * $FreeBSD$ + */ + +#ifndef _ARM_VDSO_H +#define_ARM_VDSO_H + +#defineVDSO_TIMEHANDS_MD \ + uint32_tth_res[8]; + +#endif Modified: head/sys/conf/files.arm == --- head/sys/conf/files.arm Fri Jun 22 06:44:22 2012(r237432) +++ head/sys/conf/files.arm Fri Jun 22 07:06:40 2012(r237433) @@ -79,6 +79,7 @@ libkern/umoddi3.c standard #libkern/arm/strcmp.S standard #libkern/arm/strncmp.S standard #
svn commit: r237434 - in head/lib/libc: amd64/sys gen i386/sys include sys
Author: kib
Date: Fri Jun 22 07:13:30 2012
New Revision: 237434
URL: http://svn.freebsd.org/changeset/base/237434
Log:
Use struct vdso_timehands data to implement fast gettimeofday(2) and
clock_gettime(2) functions if supported. The speedup seen in
microbenchmarks is in range 4x-7x depending on the hardware.
Only amd64 and i386 architectures are supported. Libc uses rdtsc and
kernel data to calculate current time, if enabled by kernel.
Hopefully, this code is going to migrate into vdso in some future.
Discussed with: bde
Reviewed by: jhb
Tested by:flo
MFC after:1 month
Added:
head/lib/libc/amd64/sys/__vdso_gettc.c (contents, props changed)
head/lib/libc/i386/sys/__vdso_gettc.c (contents, props changed)
head/lib/libc/sys/__vdso_gettimeofday.c (contents, props changed)
head/lib/libc/sys/clock_gettime.c (contents, props changed)
head/lib/libc/sys/gettimeofday.c (contents, props changed)
Modified:
head/lib/libc/amd64/sys/Makefile.inc
head/lib/libc/gen/aux.c
head/lib/libc/i386/sys/Makefile.inc
head/lib/libc/include/libc_private.h
head/lib/libc/sys/Makefile.inc
Modified: head/lib/libc/amd64/sys/Makefile.inc
==
--- head/lib/libc/amd64/sys/Makefile.incFri Jun 22 07:06:40 2012
(r237433)
+++ head/lib/libc/amd64/sys/Makefile.incFri Jun 22 07:13:30 2012
(r237434)
@@ -1,7 +1,8 @@
# from: Makefile.inc,v 1.1 1993/09/03 19:04:23 jtc Exp
# $FreeBSD$
-SRCS+= amd64_get_fsbase.c amd64_get_gsbase.c amd64_set_fsbase.c
amd64_set_gsbase.c
+SRCS+= amd64_get_fsbase.c amd64_get_gsbase.c amd64_set_fsbase.c \
+ amd64_set_gsbase.c __vdso_gettc.c
MDASM= vfork.S brk.S cerror.S exect.S getcontext.S pipe.S ptrace.S \
reboot.S sbrk.S setlogin.S sigreturn.S
Added: head/lib/libc/amd64/sys/__vdso_gettc.c
==
--- /dev/null 00:00:00 1970 (empty, because file is newly added)
+++ head/lib/libc/amd64/sys/__vdso_gettc.c Fri Jun 22 07:13:30 2012
(r237434)
@@ -0,0 +1,49 @@
+/*-
+ * Copyright (c) 2012 Konstantin Belousov
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *notice, this list of conditions and the following disclaimer in the
+ *documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ */
+
+#include
+__FBSDID("$FreeBSD$");
+
+#include
+#include
+#include
+#include
+
+static u_int
+__vdso_gettc_low(const struct vdso_timehands *th)
+{
+ uint32_t rv;
+
+ __asm __volatile("rdtsc; shrd %%cl, %%edx, %0"
+ : "=a" (rv) : "c" (th->th_x86_shift) : "edx");
+ return (rv);
+}
+
+u_int
+__vdso_gettc(const struct vdso_timehands *th)
+{
+
+ return (th->th_x86_shift > 0 ? __vdso_gettc_low(th) : rdtsc32());
+}
Modified: head/lib/libc/gen/aux.c
==
--- head/lib/libc/gen/aux.c Fri Jun 22 07:06:40 2012(r237433)
+++ head/lib/libc/gen/aux.c Fri Jun 22 07:13:30 2012(r237434)
@@ -66,6 +66,7 @@ __init_elf_aux_vector(void)
static pthread_once_t aux_once = PTHREAD_ONCE_INIT;
static int pagesize, osreldate, canary_len, ncpus, pagesizes_len;
static char *canary, *pagesizes;
+static void *timekeep;
static void
init_aux(void)
@@ -101,6 +102,10 @@ init_aux(void)
case AT_NCPUS:
ncpus = aux->a_un.a_val;
break;
+
+ case AT_TIMEKEEP:
+ timekeep = aux->a_un.a_ptr;
+ break;
}
}
}
@@ -163,6 +168,16 @@ _elf_aux_info(int aux, void *buf, int bu
} else
res = EINVAL;
break;
+ case AT_TIMEKEEP:
+ if (buflen == size
svn commit: r237435 - in head/sys/i386: i386 include
Author: kib
Date: Fri Jun 22 07:16:29 2012
New Revision: 237435
URL: http://svn.freebsd.org/changeset/base/237435
Log:
Enable shared page on i386, now it has a use for vdso_timehands.
MFC after:1 month
Modified:
head/sys/i386/i386/elf_machdep.c
head/sys/i386/include/vmparam.h
Modified: head/sys/i386/i386/elf_machdep.c
==
--- head/sys/i386/i386/elf_machdep.cFri Jun 22 07:13:30 2012
(r237434)
+++ head/sys/i386/i386/elf_machdep.cFri Jun 22 07:16:29 2012
(r237435)
@@ -74,12 +74,15 @@ struct sysentvec elf32_freebsd_sysvec =
.sv_setregs = exec_setregs,
.sv_fixlimit= NULL,
.sv_maxssiz = NULL,
- .sv_flags = SV_ABI_FREEBSD | SV_IA32 | SV_ILP32,
+ .sv_flags = SV_ABI_FREEBSD | SV_IA32 | SV_ILP32 | SV_SHP,
.sv_set_syscall_retval = cpu_set_syscall_retval,
.sv_fetch_syscall_args = cpu_fetch_syscall_args,
.sv_syscallnames = syscallnames,
+ .sv_shared_page_base = SHAREDPAGE,
+ .sv_shared_page_len = PAGE_SIZE,
.sv_schedtail = NULL,
};
+INIT_SYSENTVEC(elf32_sysvec, &elf32_freebsd_sysvec);
static Elf32_Brandinfo freebsd_brand_info = {
.brand = ELFOSABI_FREEBSD,
Modified: head/sys/i386/include/vmparam.h
==
--- head/sys/i386/include/vmparam.h Fri Jun 22 07:13:30 2012
(r237434)
+++ head/sys/i386/include/vmparam.h Fri Jun 22 07:16:29 2012
(r237435)
@@ -165,7 +165,8 @@
#define VM_MAXUSER_ADDRESS VADDR(PTDPTDI, 0)
-#define USRSTACK VM_MAXUSER_ADDRESS
+#defineSHAREDPAGE (VM_MAXUSER_ADDRESS - PAGE_SIZE)
+#defineUSRSTACKSHAREDPAGE
#define VM_MAX_ADDRESS VADDR(PTDPTDI, PTDPTDI)
#define VM_MIN_ADDRESS ((vm_offset_t)0)
___
svn-src-head@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
Re: svn commit: r237434 - in head/lib/libc: amd64/sys gen i386/sys include sys
On Fri, Jun 22, 2012 at 07:13:31AM +, Konstantin Belousov wrote: > Author: kib > Date: Fri Jun 22 07:13:30 2012 > New Revision: 237434 > URL: http://svn.freebsd.org/changeset/base/237434 > > Log: > Use struct vdso_timehands data to implement fast gettimeofday(2) and > clock_gettime(2) functions if supported. The speedup seen in > microbenchmarks is in range 4x-7x depending on the hardware. > > Only amd64 and i386 architectures are supported. Libc uses rdtsc and > kernel data to calculate current time, if enabled by kernel. I don't know much about x86 CPUs but is my understanding correct that TSCs are not synchronized in any way across CPUs, i.e. reading it on different CPUs may result in time going backwards etc., which is okay for this application though? Marius ___ svn-src-head@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
Re: svn commit: r237434 - in head/lib/libc: amd64/sys gen i386/sys include sys
On Fri, Jun 22, 2012 at 09:34:56AM +0200, Marius Strobl wrote: > On Fri, Jun 22, 2012 at 07:13:31AM +, Konstantin Belousov wrote: > > Author: kib > > Date: Fri Jun 22 07:13:30 2012 > > New Revision: 237434 > > URL: http://svn.freebsd.org/changeset/base/237434 > > > > Log: > > Use struct vdso_timehands data to implement fast gettimeofday(2) and > > clock_gettime(2) functions if supported. The speedup seen in > > microbenchmarks is in range 4x-7x depending on the hardware. > > > > Only amd64 and i386 architectures are supported. Libc uses rdtsc and > > kernel data to calculate current time, if enabled by kernel. > > I don't know much about x86 CPUs but is my understanding correct > that TSCs are not synchronized in any way across CPUs, i.e. > reading it on different CPUs may result in time going backwards > etc., which is okay for this application though? Generally speaking, tsc state among different CPU after boot is not synchronized, you are right. Kernel has somewhat doubtful test which verifies whether the after-boot state of tsc looks good. If the test fails, TSC is not enabled by default as timecounter, and then usermode follows kernel policy and falls back to slow syscall. So we err on the safe side. I tested this on Core i7 2xxx, where the test (usually) passes. The test we currently have fails for me at least on single-package Nehalems, where the counter should be located on uncore part. This indicates some brokeness in the code, but I did not investigated the cause. The code can be developed which adjusts tsc msrs to be in sync. Or, rtdscp instruction can be used, which allow to handle counter skew in usermode in race-free manner. While you are there. do you have comments about sparc64 TICK counter ? On SMP, the counter of BSP is used by IPI. Is it unavoidable ? pgpbUQaD5boRd.pgp Description: PGP signature
svn commit: r237436 - in head/sys/dev/cxgbe: . common firmware tom
Author: np
Date: Fri Jun 22 07:51:15 2012
New Revision: 237436
URL: http://svn.freebsd.org/changeset/base/237436
Log:
cxgbe(4): update to firmware interface 1.5.2.0; updates to shared code.
Modified:
head/sys/dev/cxgbe/common/common.h
head/sys/dev/cxgbe/common/t4_hw.c
head/sys/dev/cxgbe/common/t4_msg.h
head/sys/dev/cxgbe/firmware/t4fw_cfg.txt
head/sys/dev/cxgbe/firmware/t4fw_cfg_uwire.txt
head/sys/dev/cxgbe/firmware/t4fw_interface.h
head/sys/dev/cxgbe/osdep.h
head/sys/dev/cxgbe/t4_sge.c
head/sys/dev/cxgbe/tom/t4_cpl_io.c
Modified: head/sys/dev/cxgbe/common/common.h
==
--- head/sys/dev/cxgbe/common/common.h Fri Jun 22 07:16:29 2012
(r237435)
+++ head/sys/dev/cxgbe/common/common.h Fri Jun 22 07:51:15 2012
(r237436)
@@ -38,6 +38,8 @@ enum {
SERNUM_LEN = 24,/* Serial # length */
EC_LEN = 16,/* E/C length */
ID_LEN = 16,/* ID length */
+ PN_LEN = 16,/* Part Number length */
+ MACADDR_LEN= 12,/* MAC Address length */
};
enum { MEM_EDC0, MEM_EDC1, MEM_MC };
@@ -62,8 +64,8 @@ enum {
};
#define FW_VERSION_MAJOR 1
-#define FW_VERSION_MINOR 4
-#define FW_VERSION_MICRO 16
+#define FW_VERSION_MINOR 5
+#define FW_VERSION_MICRO 2
struct port_stats {
u64 tx_octets;/* total # of octets in good frames */
@@ -219,6 +221,8 @@ struct vpd_params {
u8 ec[EC_LEN + 1];
u8 sn[SERNUM_LEN + 1];
u8 id[ID_LEN + 1];
+ u8 pn[PN_LEN + 1];
+ u8 na[MACADDR_LEN + 1];
};
struct pci_params {
@@ -356,6 +360,8 @@ void t4_write_indirect(struct adapter *a
unsigned int data_reg, const u32 *vals,
unsigned int nregs, unsigned int start_idx);
+u32 t4_hw_pci_read_cfg4(adapter_t *adapter, int reg);
+
struct fw_filter_wr;
void t4_intr_enable(struct adapter *adapter);
@@ -374,7 +380,7 @@ int t4_seeprom_wp(struct adapter *adapte
int t4_read_flash(struct adapter *adapter, unsigned int addr, unsigned int
nwords,
u32 *data, int byte_oriented);
int t4_load_fw(struct adapter *adapter, const u8 *fw_data, unsigned int size);
-int t4_load_boot(struct adapter *adap, const u8 *boot_data,
+int t4_load_boot(struct adapter *adap, u8 *boot_data,
unsigned int boot_addr, unsigned int size);
unsigned int t4_flash_cfg_addr(struct adapter *adapter);
int t4_load_cfg(struct adapter *adapter, const u8 *cfg_data, unsigned int
size);
@@ -431,6 +437,9 @@ int t4_mem_read(struct adapter *adap, in
__be32 *data);
void t4_get_port_stats(struct adapter *adap, int idx, struct port_stats *p);
+void t4_get_port_stats_offset(struct adapter *adap, int idx,
+ struct port_stats *stats,
+ struct port_stats *offset);
void t4_get_lb_stats(struct adapter *adap, int idx, struct lb_port_stats *p);
void t4_clr_port_stats(struct adapter *adap, int idx);
@@ -472,6 +481,10 @@ int t4_fw_hello(struct adapter *adap, un
enum dev_master master, enum dev_state *state);
int t4_fw_bye(struct adapter *adap, unsigned int mbox);
int t4_fw_reset(struct adapter *adap, unsigned int mbox, int reset);
+int t4_fw_halt(struct adapter *adap, unsigned int mbox, int force);
+int t4_fw_restart(struct adapter *adap, unsigned int mbox, int reset);
+int t4_fw_upgrade(struct adapter *adap, unsigned int mbox,
+ const u8 *fw_data, unsigned int size, int force);
int t4_fw_initialize(struct adapter *adap, unsigned int mbox);
int t4_query_params(struct adapter *adap, unsigned int mbox, unsigned int pf,
unsigned int vf, unsigned int nparams, const u32 *params,
@@ -484,6 +497,10 @@ int t4_cfg_pfvf(struct adapter *adap, un
unsigned int rxqi, unsigned int rxq, unsigned int tc,
unsigned int vi, unsigned int cmask, unsigned int pmask,
unsigned int exactf, unsigned int rcaps, unsigned int wxcaps);
+int t4_alloc_vi_func(struct adapter *adap, unsigned int mbox,
+unsigned int port, unsigned int pf, unsigned int vf,
+unsigned int nmac, u8 *mac, unsigned int *rss_size,
+unsigned int portfunc, unsigned int idstype);
int t4_alloc_vi(struct adapter *adap, unsigned int mbox, unsigned int port,
unsigned int pf, unsigned int vf, unsigned int nmac, u8 *mac,
unsigned int *rss_size);
@@ -524,5 +541,10 @@ int t4_sge_ctxt_rd(struct adapter *adap,
enum ctxt_type ctype, u32 *data);
int t4_sge_ctxt_rd_bd(struct adapter *adap, unsigned int cid, enum ctxt_type
ctype,
u32 *data);
+int t4_sge_ctxt_flush(struct adapter *adap, unsigned int mbox);
int t4_handle_fw_rpl(struct adapter *adap, const __be64 *rpl);
+int t4_fwaddrspace_write(struct adapter *adap, unsigned int mbox, u32 addr,
u32 val);
+int t4_config_schedu
Re: svn commit: r237433 - in head/sys: amd64/include arm/include conf i386/include ia64/include kern mips/include pc98/include powerpc/include sparc64/include sys x86/include x86/x86
On 06/22/12 10:06, Konstantin Belousov wrote: Author: kib Date: Fri Jun 22 07:06:40 2012 New Revision: 237433 URL: http://svn.freebsd.org/changeset/base/237433 Log: Implement mechanism to export some kernel timekeeping data to usermode, using shared page. The structures and functions have vdso prefix, to indicate the intended location of the code in some future. The versioned per-algorithm data is exported in the format of struct vdso_timehands, which mostly repeats the content of in-kernel struct timehands. Usermode reading of the structure can be lockless. Compatibility export for 32bit processes on 64bit host is also provided. Kernel also provides usermode with indication about currently used timecounter, so that libc can fall back to syscall if configured timecounter is unknown to usermode code. The shared data updates are initiated both from the tc_windup(), where a fast task is queued to do the update, and from sysctl handlers which change timecounter. A manual override switch kern.timecounter.fast_gettime allows to turn off the mechanism. Only x86 architectures export the real algorithm data, and there, only for tsc timecounter. HPET counters page could be exported as well, but I prefer to not further glue the kernel and libc ABI there until proper vdso-based solution is developed. Minimal stubs neccessary for non-x86 architectures to still compile are provided. Discussed with: bde Reviewed by: jhb Tested by: flo MFC after: 1 month @@ -1360,6 +1367,7 @@ tc_windup(void) #endif timehands = th; + taskqueue_enqueue_fast(taskqueue_fast,&tc_windup_push_vdso_task); } /* Report or change the active timecounter hardware. */ This taskqueue_enqueue_fast() will schedule extra thread to run each time hardclock() fires. One thread may be not a big problem, but together with callout swi and possible other threads woken up there it will wake up several other CPU cores from sleep just to put them back in few microseconds. Now davide@ and me are trying to fix that by avoiding callout SWI use for simple tasks. Please, let's not create another problem same time. -- Alexander Motin ___ svn-src-head@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
Re: svn commit: r237434 - in head/lib/libc: amd64/sys gen i386/sys include sys
On 22 Jun 2012, at 08:34, Marius Strobl wrote: > I don't know much about x86 CPUs but is my understanding correct > that TSCs are not synchronized in any way across CPUs, i.e. > reading it on different CPUs may result in time going backwards > etc., which is okay for this application though? As long as the initial value is set on every context switch, it only matters that the TSC is monotonic and increments at an approximately constant rate. It is also possible to set the TSC value, but that's less useful in this context. The one thing to be careful about is the fact that certain power saving states will affect the speed at which the TSC increments, and so it is important to update the ticks-per-second value whenever a core goes into a low power state. This is more or less the same approach used by Xen, so most of the issues have been ironed out: Oracle complained to CPU vendors about a few corner cases and, because Oracle customers tend to buy a lot of expensive Xeon and Opteron chips, they were fixed quite promptly. David ___ svn-src-head@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
Re: svn commit: r237433 - in head/sys: amd64/include arm/include conf i386/include ia64/include kern mips/include pc98/include powerpc/include sparc64/include sys x86/include x86/x86
On Fri, Jun 22, 2012 at 11:08:50AM +0300, Alexander Motin wrote: > On 06/22/12 10:06, Konstantin Belousov wrote: > >Author: kib > >Date: Fri Jun 22 07:06:40 2012 > >New Revision: 237433 > >URL: http://svn.freebsd.org/changeset/base/237433 > > > >Log: > > Implement mechanism to export some kernel timekeeping data to > > usermode, using shared page. The structures and functions have vdso > > prefix, to indicate the intended location of the code in some future. > > > > The versioned per-algorithm data is exported in the format of struct > > vdso_timehands, which mostly repeats the content of in-kernel struct > > timehands. Usermode reading of the structure can be lockless. > > Compatibility export for 32bit processes on 64bit host is also > > provided. Kernel also provides usermode with indication about > > currently used timecounter, so that libc can fall back to syscall if > > configured timecounter is unknown to usermode code. > > > > The shared data updates are initiated both from the tc_windup(), where > > a fast task is queued to do the update, and from sysctl handlers which > > change timecounter. A manual override switch > > kern.timecounter.fast_gettime allows to turn off the mechanism. > > > > Only x86 architectures export the real algorithm data, and there, only > > for tsc timecounter. HPET counters page could be exported as well, but > > I prefer to not further glue the kernel and libc ABI there until > > proper vdso-based solution is developed. > > > > Minimal stubs neccessary for non-x86 architectures to still compile > > are provided. > > > > Discussed with: bde > > Reviewed by: jhb > > Tested by:flo > > MFC after:1 month > > > >@@ -1360,6 +1367,7 @@ tc_windup(void) > > #endif > > > > timehands = th; > >+taskqueue_enqueue_fast(taskqueue_fast,&tc_windup_push_vdso_task); > > } > > > > /* Report or change the active timecounter hardware. */ > > This taskqueue_enqueue_fast() will schedule extra thread to run each > time hardclock() fires. One thread may be not a big problem, but > together with callout swi and possible other threads woken up there it > will wake up several other CPU cores from sleep just to put them back in > few microseconds. Now davide@ and me are trying to fix that by avoiding > callout SWI use for simple tasks. Please, let's not create another > problem same time. The patch was public for quite a time. If you have some comments about it, it would be much more productive to let me know about them before the commit, not after. Anyway, what is your proposal for 'let's not create another problem same time' part of the message ? It was discussed, as a possibility, to have permanent mapping for the shared page in the KVA and to perform lock-less update of the struct vdso_timehands directly from hardclock handler. My opinion was that amount of work added by tc_windup eventhandler is not trivial, so it is better to be postponed to less critical context. It also slightly more safe to not perform lockless update for vdso_timehands, since otherwise module load which register exec handler could cause transient gettimeofday() failure in usermode. This might boil down to the fact that tc_windup function is called too often, in fact. Also, packing execution of tc_windup eventhandler together with the clock swi is fine from my POV. pgptCurkX7G0T.pgp Description: PGP signature
svn commit: r237439 - head/sys/dev/cxgbe
Author: np Date: Fri Jun 22 08:37:33 2012 New Revision: 237439 URL: http://svn.freebsd.org/changeset/base/237439 Log: Do not read registers with read side effects while performing a register dump for cxgbetool. Modified: head/sys/dev/cxgbe/t4_main.c Modified: head/sys/dev/cxgbe/t4_main.c == --- head/sys/dev/cxgbe/t4_main.cFri Jun 22 07:53:51 2012 (r237438) +++ head/sys/dev/cxgbe/t4_main.cFri Jun 22 08:37:33 2012 (r237439) @@ -2705,7 +2705,9 @@ t4_get_regs(struct adapter *sc, struct t 0xdfc0, 0xdfe0, 0xe000, 0xea7c, 0xf000, 0x11190, - 0x19040, 0x19124, + 0x19040, 0x1906c, + 0x19078, 0x19080, + 0x1908c, 0x19124, 0x19150, 0x191b0, 0x191d0, 0x191e8, 0x19238, 0x1924c, @@ -2718,49 +2720,49 @@ t4_get_regs(struct adapter *sc, struct t 0x1a190, 0x1a1c4, 0x1a1fc, 0x1a1fc, 0x1e040, 0x1e04c, - 0x1e240, 0x1e28c, + 0x1e284, 0x1e28c, 0x1e2c0, 0x1e2c0, 0x1e2e0, 0x1e2e0, 0x1e300, 0x1e384, 0x1e3c0, 0x1e3c8, 0x1e440, 0x1e44c, - 0x1e640, 0x1e68c, + 0x1e684, 0x1e68c, 0x1e6c0, 0x1e6c0, 0x1e6e0, 0x1e6e0, 0x1e700, 0x1e784, 0x1e7c0, 0x1e7c8, 0x1e840, 0x1e84c, - 0x1ea40, 0x1ea8c, + 0x1ea84, 0x1ea8c, 0x1eac0, 0x1eac0, 0x1eae0, 0x1eae0, 0x1eb00, 0x1eb84, 0x1ebc0, 0x1ebc8, 0x1ec40, 0x1ec4c, - 0x1ee40, 0x1ee8c, + 0x1ee84, 0x1ee8c, 0x1eec0, 0x1eec0, 0x1eee0, 0x1eee0, 0x1ef00, 0x1ef84, 0x1efc0, 0x1efc8, 0x1f040, 0x1f04c, - 0x1f240, 0x1f28c, + 0x1f284, 0x1f28c, 0x1f2c0, 0x1f2c0, 0x1f2e0, 0x1f2e0, 0x1f300, 0x1f384, 0x1f3c0, 0x1f3c8, 0x1f440, 0x1f44c, - 0x1f640, 0x1f68c, + 0x1f684, 0x1f68c, 0x1f6c0, 0x1f6c0, 0x1f6e0, 0x1f6e0, 0x1f700, 0x1f784, 0x1f7c0, 0x1f7c8, 0x1f840, 0x1f84c, - 0x1fa40, 0x1fa8c, + 0x1fa84, 0x1fa8c, 0x1fac0, 0x1fac0, 0x1fae0, 0x1fae0, 0x1fb00, 0x1fb84, 0x1fbc0, 0x1fbc8, 0x1fc40, 0x1fc4c, - 0x1fe40, 0x1fe8c, + 0x1fe84, 0x1fe8c, 0x1fec0, 0x1fec0, 0x1fee0, 0x1fee0, 0x1ff00, 0x1ff84, ___ svn-src-head@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
Re: svn commit: r237433 - in head/sys: amd64/include arm/include conf i386/include ia64/include kern mips/include pc98/include powerpc/include sparc64/include sys x86/include x86/x86
On 22.06.2012 11:25, Konstantin Belousov wrote: On Fri, Jun 22, 2012 at 11:08:50AM +0300, Alexander Motin wrote: On 06/22/12 10:06, Konstantin Belousov wrote: Author: kib Date: Fri Jun 22 07:06:40 2012 New Revision: 237433 URL: http://svn.freebsd.org/changeset/base/237433 Log: Implement mechanism to export some kernel timekeeping data to usermode, using shared page. The structures and functions have vdso prefix, to indicate the intended location of the code in some future. The versioned per-algorithm data is exported in the format of struct vdso_timehands, which mostly repeats the content of in-kernel struct timehands. Usermode reading of the structure can be lockless. Compatibility export for 32bit processes on 64bit host is also provided. Kernel also provides usermode with indication about currently used timecounter, so that libc can fall back to syscall if configured timecounter is unknown to usermode code. The shared data updates are initiated both from the tc_windup(), where a fast task is queued to do the update, and from sysctl handlers which change timecounter. A manual override switch kern.timecounter.fast_gettime allows to turn off the mechanism. Only x86 architectures export the real algorithm data, and there, only for tsc timecounter. HPET counters page could be exported as well, but I prefer to not further glue the kernel and libc ABI there until proper vdso-based solution is developed. Minimal stubs neccessary for non-x86 architectures to still compile are provided. Discussed with: bde Reviewed by: jhb Tested by: flo MFC after: 1 month @@ -1360,6 +1367,7 @@ tc_windup(void) #endif timehands = th; + taskqueue_enqueue_fast(taskqueue_fast,&tc_windup_push_vdso_task); } /* Report or change the active timecounter hardware. */ This taskqueue_enqueue_fast() will schedule extra thread to run each time hardclock() fires. One thread may be not a big problem, but together with callout swi and possible other threads woken up there it will wake up several other CPU cores from sleep just to put them back in few microseconds. Now davide@ and me are trying to fix that by avoiding callout SWI use for simple tasks. Please, let's not create another problem same time. The patch was public for quite a time. If you have some comments about it, it would be much more productive to let me know about them before the commit, not after. I'm sorry, I haven't seen it. My mad. Anyway, what is your proposal for 'let's not create another problem same time' part of the message ? It was discussed, as a possibility, to have permanent mapping for the shared page in the KVA and to perform lock-less update of the struct vdso_timehands directly from hardclock handler. My opinion was that amount of work added by tc_windup eventhandler is not trivial, so it is better to be postponed to less critical context. It also slightly more safe to not perform lockless update for vdso_timehands, since otherwise module load which register exec handler could cause transient gettimeofday() failure in usermode. This might boil down to the fact that tc_windup function is called too often, in fact. Also, packing execution of tc_windup eventhandler together with the clock swi is fine from my POV. I have nothing against using shared pages. On UP system I would probably have not so much against several threads. But on SMP system it will cause at least one, but in many cases two extra CPUs to be woken up. There are two or more threads to run on hardclock(): this taskqueue, callout swi and some thread(s) woken from callouts. Scheduler has no idea how heavy they are. So it will try to move each of them to separate idle CPU. Does the amount of work done in event handlers worth extra Watts consumed by rapidly waking CPUs? As quite rare person running FreeBSD on laptop, I am not sure. I am not sure even that on desktop/server this won't kill all benefit of fast clocks by limiting TurboBoost. -- Alexander Motin ___ svn-src-head@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
Re: svn commit: r237412 - in head: sys/contrib/dev/acpica sys/contrib/dev/acpica/common sys/contrib/dev/acpica/compiler sys/contrib/dev/acpica/components/debugger sys/contrib/dev/acpica/components/dis
On Fri, Jun 22, 2012 at 12:40:45AM +, Jung-uk Kim wrote: > Author: jkim > Date: Fri Jun 22 00:40:44 2012 > New Revision: 237412 > URL: http://svn.freebsd.org/changeset/base/237412 > > Log: > Merge ACPICA 20120620. I think it is caused by import, malloc is called while spinlock is held: ACPI: Executed 1 blocks of module-level executable AML code acpi0: Power Button (fixed) panic: blockable sleep lock (sleep mutex) 16 @ /usr/home/kostik/work/build/bsd/DEV/src/sys/vm/uma_core.c:2040 cpuid = 0 KDB: stack backtrace: db_trace_self_wrapper() at 0x80279d0a = db_trace_self_wrapper+0x2a panic() at 0x80325a28 = panic+0x1d8 witness_checkorder() at 0x80377938 = witness_checkorder+0x1d8 _mtx_lock_flags() at 0x80313777 = _mtx_lock_flags+0x87 uma_zalloc_arg() at 0x804db97a = uma_zalloc_arg+0x2ea malloc() at 0x8030f7cd = malloc+0xbd AcpiUtAllocate() at 0x802704f1 = AcpiUtAllocate+0x31 AcpiUtAllocateZeroed() at 0x80270559 = AcpiUtAllocateZeroed+0x19 AcpiSetupGpeForWake() at 0x8025becc = AcpiSetupGpeForWake+0xfc acpi_probe_child() at 0x802848e9 = acpi_probe_child+0x2b9 AcpiNsWalkNamespace() at 0x802686ef = AcpiNsWalkNamespace+0x15f AcpiWalkNamespace() at 0x80268c0f = AcpiWalkNamespace+0xbf acpi_attach() at 0x80283e43 = acpi_attach+0x993 device_attach() at 0x8035ab62 = device_attach+0x72 bus_generic_attach() at 0x8035bd3a = bus_generic_attach+0x1a nexus_acpi_attach() at 0x804fbd89 = nexus_acpi_attach+0x69 device_attach() at 0x8035ab62 = device_attach+0x72 bus_generic_new_pass() at 0x8035bf46 = bus_generic_new_pass+0xd6 bus_set_pass() at 0x803597ba = bus_set_pass+0x7a configure() at 0x804fd51a = configure+0xa mi_startup() at 0x802d5c17 = mi_startup+0x77 btext() at 0x802531cc = btext+0x2c KDB: enter: panic [ thread pid 0 tid 10 ] Stopped at 0x8035fffb = kdb_enter+0x3b:movq $0,0x1027d62(%rip) db> show alllocks Process 0 (kernel) thread 0x80837cc0 (10) exclusive sleep mutex Giant (Giant) r = 0 (0x8133a900) locked @ /usr/home/kostik/work/build/bsd/DEV/src/sys/kern/kern_module.c:116 exclusive spin mutex ACPI lock (0xfe000321da80) (ACPI lock (0xfe000321da80)) r = 0 (0xfe000321da80) locked @ /usr/home/kostik/work/build/bsd/DEV/src/sys/dev/acpica/Osd/OsdSynch.c:535 pgp6HExkXhokZ.pgp Description: PGP signature
Re: svn commit: r237433 - in head/sys: amd64/include arm/include conf i386/include ia64/include kern mips/include pc98/include powerpc/include sparc64/include sys x86/include x86/x86
On Fri, Jun 22, 2012 at 11:54:28AM +0300, Alexander Motin wrote: > On 22.06.2012 11:25, Konstantin Belousov wrote: > >On Fri, Jun 22, 2012 at 11:08:50AM +0300, Alexander Motin wrote: > >>On 06/22/12 10:06, Konstantin Belousov wrote: > >>>Author: kib > >>>Date: Fri Jun 22 07:06:40 2012 > >>>New Revision: 237433 > >>>URL: http://svn.freebsd.org/changeset/base/237433 > >>> > >>>Log: > >>> Implement mechanism to export some kernel timekeeping data to > >>> usermode, using shared page. The structures and functions have vdso > >>> prefix, to indicate the intended location of the code in some future. > >>> > >>> The versioned per-algorithm data is exported in the format of struct > >>> vdso_timehands, which mostly repeats the content of in-kernel struct > >>> timehands. Usermode reading of the structure can be lockless. > >>> Compatibility export for 32bit processes on 64bit host is also > >>> provided. Kernel also provides usermode with indication about > >>> currently used timecounter, so that libc can fall back to syscall if > >>> configured timecounter is unknown to usermode code. > >>> > >>> The shared data updates are initiated both from the tc_windup(), where > >>> a fast task is queued to do the update, and from sysctl handlers which > >>> change timecounter. A manual override switch > >>> kern.timecounter.fast_gettime allows to turn off the mechanism. > >>> > >>> Only x86 architectures export the real algorithm data, and there, only > >>> for tsc timecounter. HPET counters page could be exported as well, but > >>> I prefer to not further glue the kernel and libc ABI there until > >>> proper vdso-based solution is developed. > >>> > >>> Minimal stubs neccessary for non-x86 architectures to still compile > >>> are provided. > >>> > >>> Discussed with: bde > >>> Reviewed by:jhb > >>> Tested by: flo > >>> MFC after: 1 month > >> > >> > >>>@@ -1360,6 +1367,7 @@ tc_windup(void) > >>> #endif > >>> > >>> timehands = th; > >>>+ taskqueue_enqueue_fast(taskqueue_fast,&tc_windup_push_vdso_task); > >>> } > >>> > >>> /* Report or change the active timecounter hardware. */ > >> > >>This taskqueue_enqueue_fast() will schedule extra thread to run each > >>time hardclock() fires. One thread may be not a big problem, but > >>together with callout swi and possible other threads woken up there it > >>will wake up several other CPU cores from sleep just to put them back in > >>few microseconds. Now davide@ and me are trying to fix that by avoiding > >>callout SWI use for simple tasks. Please, let's not create another > >>problem same time. > > > >The patch was public for quite a time. If you have some comments about > >it, it would be much more productive to let me know about them before > >the commit, not after. > > I'm sorry, I haven't seen it. My mad. > > >Anyway, what is your proposal for 'let's not create another problem > >same time' part of the message ? It was discussed, as a possibility, > >to have permanent mapping for the shared page in the KVA and to perform > >lock-less update of the struct vdso_timehands directly from hardclock > >handler. My opinion was that amount of work added by tc_windup > >eventhandler is not trivial, so it is better to be postponed to > >less critical context. It also slightly more safe to not perform > >lockless update for vdso_timehands, since otherwise module load which > >register exec handler could cause transient gettimeofday() failure > >in usermode. > > > >This might boil down to the fact that tc_windup function is called > >too often, in fact. Also, packing execution of tc_windup eventhandler > >together with the clock swi is fine from my POV. > > I have nothing against using shared pages. On UP system I would probably > have not so much against several threads. But on SMP system it will > cause at least one, but in many cases two extra CPUs to be woken up. > There are two or more threads to run on hardclock(): this taskqueue, > callout swi and some thread(s) woken from callouts. Scheduler has no > idea how heavy they are. So it will try to move each of them to separate > idle CPU. Does the amount of work done in event handlers worth extra > Watts consumed by rapidly waking CPUs? As quite rare person running > FreeBSD on laptop, I am not sure. I am not sure even that on > desktop/server this won't kill all benefit of fast clocks by limiting > TurboBoost. Patch below would probably work, but I cannot test it right now on real hardware due to ACPI issue. It worked for me in qemu. commit 4f2ffd93b36d20eae61495776fc6b0855745fd7f Author: Konstantin Belousov Date: Fri Jun 22 13:19:22 2012 +0300 Use persistent kernel mapping of the shared page, and update the vdso_timehands from hardclock, instead of scheduling task. diff --git a/sys/kern/kern_exec.c b/sys/kern/kern_exec.c index 80502e3..9365223 100644 --- a/sys/kern/kern_exec.c +++ b/sys/kern/kern_exec.c @@ -1517,42 +1517,13 @@ exec_unregis
Re: svn commit: r237433 - in head/sys: amd64/include arm/include conf i386/include ia64/include kern mips/include pc98/include powerpc/include sparc64/include sys x86/include x86/x86
On Fri, Jun 22, 2012 at 01:23:42PM +0300, Konstantin Belousov wrote: > On Fri, Jun 22, 2012 at 11:54:28AM +0300, Alexander Motin wrote: > > On 22.06.2012 11:25, Konstantin Belousov wrote: > > >On Fri, Jun 22, 2012 at 11:08:50AM +0300, Alexander Motin wrote: > > >>On 06/22/12 10:06, Konstantin Belousov wrote: > > >>>Author: kib > > >>>Date: Fri Jun 22 07:06:40 2012 > > >>>New Revision: 237433 > > >>>URL: http://svn.freebsd.org/changeset/base/237433 > > >>> > > >>>Log: > > >>> Implement mechanism to export some kernel timekeeping data to > > >>> usermode, using shared page. The structures and functions have vdso > > >>> prefix, to indicate the intended location of the code in some future. > > >>> > > >>> The versioned per-algorithm data is exported in the format of struct > > >>> vdso_timehands, which mostly repeats the content of in-kernel struct > > >>> timehands. Usermode reading of the structure can be lockless. > > >>> Compatibility export for 32bit processes on 64bit host is also > > >>> provided. Kernel also provides usermode with indication about > > >>> currently used timecounter, so that libc can fall back to syscall if > > >>> configured timecounter is unknown to usermode code. > > >>> > > >>> The shared data updates are initiated both from the tc_windup(), where > > >>> a fast task is queued to do the update, and from sysctl handlers which > > >>> change timecounter. A manual override switch > > >>> kern.timecounter.fast_gettime allows to turn off the mechanism. > > >>> > > >>> Only x86 architectures export the real algorithm data, and there, only > > >>> for tsc timecounter. HPET counters page could be exported as well, but > > >>> I prefer to not further glue the kernel and libc ABI there until > > >>> proper vdso-based solution is developed. > > >>> > > >>> Minimal stubs neccessary for non-x86 architectures to still compile > > >>> are provided. > > >>> > > >>> Discussed with: bde > > >>> Reviewed by: jhb > > >>> Tested by:flo > > >>> MFC after:1 month > > >> > > >> > > >>>@@ -1360,6 +1367,7 @@ tc_windup(void) > > >>> #endif > > >>> > > >>> timehands = th; > > >>>+ > > >>>taskqueue_enqueue_fast(taskqueue_fast,&tc_windup_push_vdso_task); > > >>> } > > >>> > > >>> /* Report or change the active timecounter hardware. */ > > >> > > >>This taskqueue_enqueue_fast() will schedule extra thread to run each > > >>time hardclock() fires. One thread may be not a big problem, but > > >>together with callout swi and possible other threads woken up there it > > >>will wake up several other CPU cores from sleep just to put them back in > > >>few microseconds. Now davide@ and me are trying to fix that by avoiding > > >>callout SWI use for simple tasks. Please, let's not create another > > >>problem same time. > > > > > >The patch was public for quite a time. If you have some comments about > > >it, it would be much more productive to let me know about them before > > >the commit, not after. > > > > I'm sorry, I haven't seen it. My mad. > > > > >Anyway, what is your proposal for 'let's not create another problem > > >same time' part of the message ? It was discussed, as a possibility, > > >to have permanent mapping for the shared page in the KVA and to perform > > >lock-less update of the struct vdso_timehands directly from hardclock > > >handler. My opinion was that amount of work added by tc_windup > > >eventhandler is not trivial, so it is better to be postponed to > > >less critical context. It also slightly more safe to not perform > > >lockless update for vdso_timehands, since otherwise module load which > > >register exec handler could cause transient gettimeofday() failure > > >in usermode. > > > > > >This might boil down to the fact that tc_windup function is called > > >too often, in fact. Also, packing execution of tc_windup eventhandler > > >together with the clock swi is fine from my POV. > > > > I have nothing against using shared pages. On UP system I would probably > > have not so much against several threads. But on SMP system it will > > cause at least one, but in many cases two extra CPUs to be woken up. > > There are two or more threads to run on hardclock(): this taskqueue, > > callout swi and some thread(s) woken from callouts. Scheduler has no > > idea how heavy they are. So it will try to move each of them to separate > > idle CPU. Does the amount of work done in event handlers worth extra > > Watts consumed by rapidly waking CPUs? As quite rare person running > > FreeBSD on laptop, I am not sure. I am not sure even that on > > desktop/server this won't kill all benefit of fast clocks by limiting > > TurboBoost. > > Patch below would probably work, but I cannot test it right now on real > hardware due to ACPI issue. It worked for me in qemu. > > commit 4f2ffd93b36d20eae61495776fc6b0855745fd7f > Author: Konstantin Belousov > Date: Fri Jun 22 13:19:22 2012 +0300 > > Use persistent kern
Re: svn commit: r237433 - in head/sys: amd64/include arm/include conf i386/include ia64/include kern mips/include pc98/include powerpc/include sparc64/include sys x86/include x86/x86
On 06/22/12 13:46, Konstantin Belousov wrote: On Fri, Jun 22, 2012 at 01:23:42PM +0300, Konstantin Belousov wrote: On Fri, Jun 22, 2012 at 11:54:28AM +0300, Alexander Motin wrote: On 22.06.2012 11:25, Konstantin Belousov wrote: On Fri, Jun 22, 2012 at 11:08:50AM +0300, Alexander Motin wrote: On 06/22/12 10:06, Konstantin Belousov wrote: Author: kib Date: Fri Jun 22 07:06:40 2012 New Revision: 237433 URL: http://svn.freebsd.org/changeset/base/237433 Log: Implement mechanism to export some kernel timekeeping data to usermode, using shared page. The structures and functions have vdso prefix, to indicate the intended location of the code in some future. The versioned per-algorithm data is exported in the format of struct vdso_timehands, which mostly repeats the content of in-kernel struct timehands. Usermode reading of the structure can be lockless. Compatibility export for 32bit processes on 64bit host is also provided. Kernel also provides usermode with indication about currently used timecounter, so that libc can fall back to syscall if configured timecounter is unknown to usermode code. The shared data updates are initiated both from the tc_windup(), where a fast task is queued to do the update, and from sysctl handlers which change timecounter. A manual override switch kern.timecounter.fast_gettime allows to turn off the mechanism. Only x86 architectures export the real algorithm data, and there, only for tsc timecounter. HPET counters page could be exported as well, but I prefer to not further glue the kernel and libc ABI there until proper vdso-based solution is developed. Minimal stubs neccessary for non-x86 architectures to still compile are provided. Discussed with: bde Reviewed by: jhb Tested by: flo MFC after: 1 month @@ -1360,6 +1367,7 @@ tc_windup(void) #endif timehands = th; + taskqueue_enqueue_fast(taskqueue_fast,&tc_windup_push_vdso_task); } /* Report or change the active timecounter hardware. */ This taskqueue_enqueue_fast() will schedule extra thread to run each time hardclock() fires. One thread may be not a big problem, but together with callout swi and possible other threads woken up there it will wake up several other CPU cores from sleep just to put them back in few microseconds. Now davide@ and me are trying to fix that by avoiding callout SWI use for simple tasks. Please, let's not create another problem same time. The patch was public for quite a time. If you have some comments about it, it would be much more productive to let me know about them before the commit, not after. I'm sorry, I haven't seen it. My mad. Anyway, what is your proposal for 'let's not create another problem same time' part of the message ? It was discussed, as a possibility, to have permanent mapping for the shared page in the KVA and to perform lock-less update of the struct vdso_timehands directly from hardclock handler. My opinion was that amount of work added by tc_windup eventhandler is not trivial, so it is better to be postponed to less critical context. It also slightly more safe to not perform lockless update for vdso_timehands, since otherwise module load which register exec handler could cause transient gettimeofday() failure in usermode. This might boil down to the fact that tc_windup function is called too often, in fact. Also, packing execution of tc_windup eventhandler together with the clock swi is fine from my POV. I have nothing against using shared pages. On UP system I would probably have not so much against several threads. But on SMP system it will cause at least one, but in many cases two extra CPUs to be woken up. There are two or more threads to run on hardclock(): this taskqueue, callout swi and some thread(s) woken from callouts. Scheduler has no idea how heavy they are. So it will try to move each of them to separate idle CPU. Does the amount of work done in event handlers worth extra Watts consumed by rapidly waking CPUs? As quite rare person running FreeBSD on laptop, I am not sure. I am not sure even that on desktop/server this won't kill all benefit of fast clocks by limiting TurboBoost. Patch below would probably work, but I cannot test it right now on real hardware due to ACPI issue. It worked for me in qemu. commit 4f2ffd93b36d20eae61495776fc6b0855745fd7f Author: Konstantin Belousov Date: Fri Jun 22 13:19:22 2012 +0300 Use persistent kernel mapping of the shared page, and update the vdso_timehands from hardclock, instead of scheduling task. Slightly improved version. Since tc_fill_vdso_timehands is now called from hardclock context, thee is no need to spin waiting for valid current generation of timehands. I can't evaluate how much "hackish" this is from memory management side, but I'm glad there is some viable solution. Thank you! -- Alexander Motin ___ svn-src-head@freebsd.org m
Re: svn commit: r237434 - in head/lib/libc: amd64/sys gen i386/sys include sys
On Friday, June 22, 2012 3:34:56 am Marius Strobl wrote: > On Fri, Jun 22, 2012 at 07:13:31AM +, Konstantin Belousov wrote: > > Author: kib > > Date: Fri Jun 22 07:13:30 2012 > > New Revision: 237434 > > URL: http://svn.freebsd.org/changeset/base/237434 > > > > Log: > > Use struct vdso_timehands data to implement fast gettimeofday(2) and > > clock_gettime(2) functions if supported. The speedup seen in > > microbenchmarks is in range 4x-7x depending on the hardware. > > > > Only amd64 and i386 architectures are supported. Libc uses rdtsc and > > kernel data to calculate current time, if enabled by kernel. > > I don't know much about x86 CPUs but is my understanding correct > that TSCs are not synchronized in any way across CPUs, i.e. > reading it on different CPUs may result in time going backwards > etc., which is okay for this application though? Hmm, in practice I have found that on modern x86 CPUs (Penryn and later) the TSC is in fact sychronized across packages at work. At least, when I measure skew across packages it appears to be consistent with the time it would take a write to propagate from one to the other. -- John Baldwin ___ svn-src-head@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
Re: svn commit: r237412 - in head: sys/contrib/dev/acpica sys/contrib/dev/acpica/common sys/contrib/dev/acpica/compiler sys/contrib/dev/acpica/components/debugger sys/contrib/dev/acpica/components/dis
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 2012-06-22 05:58:47 -0400, Konstantin Belousov wrote: > On Fri, Jun 22, 2012 at 12:40:45AM +, Jung-uk Kim wrote: >> Author: jkim Date: Fri Jun 22 00:40:44 2012 New Revision: 237412 >> URL: http://svn.freebsd.org/changeset/base/237412 >> >> Log: Merge ACPICA 20120620. > > I think it is caused by import, malloc is called while spinlock is > held: > > ACPI: Executed 1 blocks of module-level executable AML code acpi0: > Power Button (fixed) panic: blockable sleep lock (sleep mutex) 16 @ > /usr/home/kostik/work/build/bsd/DEV/src/sys/vm/uma_core.c:2040 > cpuid = 0 KDB: stack backtrace: db_trace_self_wrapper() at > 0x80279d0a = db_trace_self_wrapper+0x2a panic() at > 0x80325a28 = panic+0x1d8 witness_checkorder() at > 0x80377938 = witness_checkorder+0x1d8 _mtx_lock_flags() at > 0x80313777 = _mtx_lock_flags+0x87 uma_zalloc_arg() at > 0x804db97a = uma_zalloc_arg+0x2ea malloc() at > 0x8030f7cd = malloc+0xbd AcpiUtAllocate() at > 0x802704f1 = AcpiUtAllocate+0x31 AcpiUtAllocateZeroed() at > 0x80270559 = AcpiUtAllocateZeroed+0x19 > AcpiSetupGpeForWake() at 0x8025becc = > AcpiSetupGpeForWake+0xfc acpi_probe_child() at 0x802848e9 = > acpi_probe_child+0x2b9 AcpiNsWalkNamespace() at 0x802686ef > = AcpiNsWalkNamespace+0x15f AcpiWalkNamespace() at > 0x80268c0f = AcpiWalkNamespace+0xbf acpi_attach() at > 0x80283e43 = acpi_attach+0x993 device_attach() at > 0x8035ab62 = device_attach+0x72 bus_generic_attach() at > 0x8035bd3a = bus_generic_attach+0x1a nexus_acpi_attach() at > 0x804fbd89 = nexus_acpi_attach+0x69 device_attach() at > 0x8035ab62 = device_attach+0x72 bus_generic_new_pass() at > 0x8035bf46 = bus_generic_new_pass+0xd6 bus_set_pass() at > 0x803597ba = bus_set_pass+0x7a configure() at > 0x804fd51a = configure+0xa mi_startup() at > 0x802d5c17 = mi_startup+0x77 btext() at 0x802531cc > = btext+0x2c KDB: enter: panic [ thread pid 0 tid 10 ] Stopped > at 0x8035fffb = kdb_enter+0x3b:movq > $0,0x1027d62(%rip) db> show alllocks Process 0 (kernel) thread > 0x80837cc0 (10) exclusive sleep mutex Giant (Giant) r = > 0 (0x8133a900) locked @ > /usr/home/kostik/work/build/bsd/DEV/src/sys/kern/kern_module.c:116 > exclusive spin mutex ACPI lock (0xfe000321da80) (ACPI lock > (0xfe000321da80)) r = 0 (0xfe000321da80) locked @ > /usr/home/kostik/work/build/bsd/DEV/src/sys/dev/acpica/Osd/OsdSynch.c:535 Hmm... > I was afraid it might happen. I'll look into it ASAP. Sorry for the trouble, Jung-uk Kim -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.19 (FreeBSD) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAk/klPoACgkQmlay1b9qnVMg+QCggRXxXE2Wpc6kXWSpwbykhrWq RvQAniVjs7p/084ZnlDPA+GQqyxTIF/w =+7Iw -END PGP SIGNATURE- ___ svn-src-head@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r237445 - head/sys/i386/i386
Author: kib
Date: Fri Jun 22 16:05:56 2012
New Revision: 237445
URL: http://svn.freebsd.org/changeset/base/237445
Log:
Commit changes missed from r237435. Properly calculate the signal
trampoline addresses after the shared page is enabled. Handle FreeBSD
ABIs without shared page support too.
Reported and tested by: David Wolfskill
(previous version)
Pointy hat to: kib
MFC after: 1 month
Modified:
head/sys/i386/i386/machdep.c
Modified: head/sys/i386/i386/machdep.c
==
--- head/sys/i386/i386/machdep.cFri Jun 22 14:58:40 2012
(r237444)
+++ head/sys/i386/i386/machdep.cFri Jun 22 16:05:56 2012
(r237445)
@@ -469,7 +469,13 @@ osendsig(sig_t catcher, ksiginfo_t *ksi,
}
regs->tf_esp = (int)fp;
- regs->tf_eip = PS_STRINGS - szosigcode;
+ if (p->p_sysent->sv_sigcode_base != 0) {
+ regs->tf_eip = p->p_sysent->sv_sigcode_base + szsigcode -
+ szosigcode;
+ } else {
+ /* a.out sysentvec does not use shared page */
+ regs->tf_eip = p->p_sysent->sv_psstrings - szosigcode;
+ }
regs->tf_eflags &= ~(PSL_T | PSL_D);
regs->tf_cs = _ucodesel;
regs->tf_ds = _udatasel;
@@ -596,7 +602,8 @@ freebsd4_sendsig(sig_t catcher, ksiginfo
}
regs->tf_esp = (int)sfp;
- regs->tf_eip = PS_STRINGS - szfreebsd4_sigcode;
+ regs->tf_eip = p->p_sysent->sv_sigcode_base + szsigcode -
+ szfreebsd4_sigcode;
regs->tf_eflags &= ~(PSL_T | PSL_D);
regs->tf_cs = _ucodesel;
regs->tf_ds = _udatasel;
@@ -747,7 +754,7 @@ sendsig(sig_t catcher, ksiginfo_t *ksi,
}
regs->tf_esp = (int)sfp;
- regs->tf_eip = PS_STRINGS - *(p->p_sysent->sv_szsigcode);
+ regs->tf_eip = p->p_sysent->sv_sigcode_base;
regs->tf_eflags &= ~(PSL_T | PSL_D);
regs->tf_cs = _ucodesel;
regs->tf_ds = _udatasel;
___
svn-src-head@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r237446 - head/sys/cam
Author: mav
Date: Fri Jun 22 16:20:13 2012
New Revision: 237446
URL: http://svn.freebsd.org/changeset/base/237446
Log:
Don't print SCSI Queue Full and CAM_REQUEUE_REQ statuses as errors if they
were handled and retried. They are part of normal operation for SCSI TCQ.
MFC after:3 days
Modified:
head/sys/cam/cam_periph.c
Modified: head/sys/cam/cam_periph.c
==
--- head/sys/cam/cam_periph.c Fri Jun 22 16:05:56 2012(r237445)
+++ head/sys/cam/cam_periph.c Fri Jun 22 16:20:13 2012(r237446)
@@ -1354,6 +1354,7 @@ camperiphscsistatuserror(union ccb *ccb,
}
*timeout = 0;
error = ERESTART;
+ *print = 0;
break;
}
/* FALLTHROUGH */
@@ -1683,8 +1684,10 @@ cam_periph_error(union ccb *ccb, cam_fla
} else if (sense_flags & SF_NO_RETRY) {
error = EIO;
action_string = "Retry was blocked";
- } else
+ } else {
error = ERESTART;
+ print = 0;
+ }
break;
case CAM_RESRC_UNAVAIL:
/* Wait a bit for the resource shortage to abate. */
___
svn-src-head@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r237448 - head/lib/libedit
Author: pfg Date: Fri Jun 22 18:01:22 2012 New Revision: 237448 URL: http://svn.freebsd.org/changeset/base/237448 Log: Merge changes from upstream libedit. Our libedit has been diverging from the mainstream version maintained in NetBSD. As a consequence it has been difficult to do an appropriate MFV and we have been bringing only partial updates. Here we update most of the files to at least match the version available in NetBSD's snapshot of 20091228. This version was chosen because it still doesn't include wide character support (UTF-8), which involves many changes and new files. From NetBSD's logs: Dec 15 22:13:33 2006 - editline.3 el.c el.h histedit.h add EL_GETFP, and EL_SETFP. Apr 5 15:53:28 2008 - editline.3 el.c histedit.h readline.c add EL_REFRESH for the benefit of readline Sep 10 15:45:37 2008 - common.c el.c read.c refresh.c sig.c term.c term.h tty.c Allow a single process to control multiple ttys (for pthreads using _REENTRANT) using multiple EditLine objects. Jan 18 12:17:24 2009 - el.c read.c readline.c fix -Wsign-compare issues Feb 6 14:40:32 2009 - history.c Plug memory leak, from MySQL. Feb 5 19:15:44 2009 - histedit.h read.c match documentation in el_push Feb 6 13:14:37 2009 - vi.c Portability fix. Feb 12 13:39:49 2009 - readline.c term.c More fixes for existing portability stuff. Feb 15 21:24:13 2009 - el.h read.c don't restart on EINTR, instead return NULL immediately. From Anon Ymous Feb 15 21:25:01 2009 - sig.c sig.h in order for read() to return EINTR we need to use sigaction, not signal, otherwise SA_RESTART is set. Feb 15 21:55:23 2009 - chared.c chared.h common.c emacs.c filecomplete.c filecomplete.h key.c key.h read.c readline.c refresh.c search.c term.c tokenizer.c tty.c vi.c pass lint on _LP64. Feb 17 21:34:26 2009 - el.c histedit.hprompt.c prompt.h allow for a prompt argument. Feb 18 15:04:40 2009 - sig.c SA_RESTART for all signals but SIGINT. From Anon Ymous. Feb 19 15:20:22 2009 - read.c sig.c sig.h reset and redraw on sigcont. From Anon Ymous. Feb 21 23:31:56 2009 - key.c key.h readline.c vi.c more size_t stuff. Mar 10 20:46:15 2009 - editline.3 read.c make el_gets set the count to -1 on error to distinguish between EOF and error. Mar 31 17:38:27 2009 - editline.3 el.c histedit.h prompt.c prompt.h refresh.c term.c term.h Implement literal prompt sequences. Now someone can implement RL_PROMPT_START_LITERAL/RL_PROMPT_END_LITERAL :-) Mar 31 21:33:17 2009 - term.c cast to size_t to avoid sign / unsigned comparison warning. Apr 23 02:03 2009 - term.c Apply patch (requested by msaitoh in ticket #2007): Coverity CID 1668: Plug memory leak when malloc() failed.:55 2009 May 11 18:33:30 2009 - editline.3 el.c histedit.h restore binary compatibility by providing new prompt functions that take an extra literal character. May 19 21:45:14 2009 - refresh.c always scroll when we advance past bottom. From Caleb Welton cwelton at greenplum dot com. Jul 17 12:27:57 2009 - term.c - off by one in the term.h case. - make code more similar to tcsh (if we want to handle wide chars, this is needed; for now it is a no-op) Jul 22 15:56:29 2009 - el.c Move filename to the scope it is being used. From Michael Cook mcook at bbn dot com Jul 22 15:57:00 2009 - read.c Always initialize nread since it is an out param. From Michael Cook mcook at bbn dot com Jul 22 18:25:26 2009 - el.c Only need path if we have issetugid... From Anon Ymous Jul 25 21:19:23 2009 - el.c Ignore comment lines in .editrc from Jess Thrysoee Sep 7 21:24:33 2009 histedit.h history.c readline.c apply apple patches from: http://opensource.apple.com/source/libedit/libedit-11/patches/ Dec 28 21:52:43 2009 - refresh.c Fix bug where tab completion on the second or > line that caused listing ended up corrupting the display by an extra space in the beginning. Reported by Mac Chan. Dec 28 22:15:36 2009 - refresh.c term.c reduce diff with tcsh Obtained from:NetBSD Tested by:bapt, jilles and current@ MFC after:1 week Modified: head/lib/libedit/common.c head/lib/libedit/editline.3 head/lib/libedit/editrc.5 head/lib/libedit/el.c head/lib/libedit/el.h head/lib/libedit/histedit.h head/lib/libedit/history.c head/lib/libedit/key.c head/lib/libedit/key.h head/lib/libedit/prompt.c head/lib/libedit/prompt.h head/lib/libedit/read.c head/lib/libedit/refresh.c head/lib/libedit/search.c head/lib/libedit/sig.c head/lib/libedit/sig.h head/lib/libedit/term.c head/lib/libedit/term.h head/lib/libedit/tty.c head/lib/libedit/vi.c Modified: head/lib/libedit/common.c == --- head/lib/libedit/common.c Fri Jun 22 16:31:00 2012(r237447) +++ he
svn commit: r237451 - head/sys/vm
Author: attilio
Date: Fri Jun 22 18:34:11 2012
New Revision: 237451
URL: http://svn.freebsd.org/changeset/base/237451
Log:
- Add a comment explaining the locking of the cached pages pool held
by vm_objects.
- Add flags for the per-object lock and free pages queue mutex lock.
Use the newly added flags to mark the cache root within the vm_object
structure.
Please note that other vm_object members should be marked with correct
locking but they are left for other commits.
In collabouration with: alc
MFC after:3 days3 days3 days
Modified:
head/sys/vm/vm_object.h
Modified: head/sys/vm/vm_object.h
==
--- head/sys/vm/vm_object.h Fri Jun 22 18:20:26 2012(r237450)
+++ head/sys/vm/vm_object.h Fri Jun 22 18:34:11 2012(r237451)
@@ -76,8 +76,21 @@
*
* vm_object_t Virtual memory object.
*
+ * The root of cached pages pool is protected by both the per-object mutex
+ * and the free pages queue mutex.
+ * On insert in the cache splay tree, the per-object mutex is expected
+ * to be already held and the free pages queue mutex will be
+ * acquired during the operation too.
+ * On remove and lookup from the cache splay tree, only the free
+ * pages queue mutex is expected to be locked.
+ * These rules allow for reliably checking for the presence of cached
+ * pages with only the per-object lock held, thereby reducing contention
+ * for the free pages queue mutex.
+ *
* List of locks
* (c) const until freed
+ * (o) per-object mutex
+ * (f) free pages queue mutex
*
*/
@@ -102,7 +115,7 @@ struct vm_object {
vm_ooffset_t backing_object_offset;/* Offset in backing object */
TAILQ_ENTRY(vm_object) pager_object_list; /* list of all objects of
this pager type */
LIST_HEAD(, vm_reserv) rvq; /* list of reservations */
- vm_page_t cache;/* root of the cache page splay tree */
+ vm_page_t cache;/* (o + f) root of the cache page splay
tree */
void *handle;
union {
/*
___
svn-src-head@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r237452 - in head: sbin/camcontrol sys/cam/scsi
Author: ken
Date: Fri Jun 22 18:57:06 2012
New Revision: 237452
URL: http://svn.freebsd.org/changeset/base/237452
Log:
Change 'camcontrol defects' to first probe a drive to find out how much
defect information it has before grabbing the full defect list.
This works around a bug with some Hitachi drives that generate data overrun
errors when they are asked for more defect data than they have.
The change is done in a spec-compliant way, so it should have no negative
impact on drives that don't have this issue.
This is based on work originally done at Sandvine.
scsi_da.h:Add a define for the maximum amount of data that can be
contained in a defect list.
camcontrol.c: Update the readdefects() function to issue an initial
command to determine the length of the defect list, and
then use that length in the request for the full defect
list.
camcontrol.8: Add a note that some drives will report 0 defects available
if you don't request either the PLIST or GLIST.
Submitted by: Mark Johnston (original version)
MFC after:3 days
Modified:
head/sbin/camcontrol/camcontrol.8
head/sbin/camcontrol/camcontrol.c
head/sys/cam/scsi/scsi_da.h
Modified: head/sbin/camcontrol/camcontrol.8
==
--- head/sbin/camcontrol/camcontrol.8 Fri Jun 22 18:34:11 2012
(r237451)
+++ head/sbin/camcontrol/camcontrol.8 Fri Jun 22 18:57:06 2012
(r237452)
@@ -497,6 +497,8 @@ is specified,
.Nm
will print out the number of defects given in the READ DEFECT DATA header
returned from the drive.
+Some drives will report 0 defects if neither the primary or grown defect
+lists are requested.
.It Ic modepage
Allows the user to display and optionally edit a SCSI mode page.
The mode
Modified: head/sbin/camcontrol/camcontrol.c
==
--- head/sbin/camcontrol/camcontrol.c Fri Jun 22 18:34:11 2012
(r237451)
+++ head/sbin/camcontrol/camcontrol.c Fri Jun 22 18:57:06 2012
(r237452)
@@ -1799,13 +1799,14 @@ readdefects(struct cam_device *device, i
union ccb *ccb = NULL;
struct scsi_read_defect_data_10 *rdd_cdb;
u_int8_t *defect_list = NULL;
- u_int32_t dlist_length = 65000;
+ u_int32_t max_dlist_length = SRDD10_MAX_LENGTH, dlist_length = 0;
u_int32_t returned_length = 0;
u_int32_t num_returned = 0;
u_int8_t returned_format;
unsigned int i;
int c, error = 0;
- int lists_specified = 0;
+ int lists_specified;
+ int get_length = 1;
while ((c = getopt(argc, argv, combinedopt)) != -1) {
switch(c){
@@ -1842,20 +1843,33 @@ readdefects(struct cam_device *device, i
ccb = cam_getccb(device);
/*
-* Hopefully 65000 bytes is enough to hold the defect list. If it
-* isn't, the disk is probably dead already. We'd have to go with
-* 12 byte command (i.e. alloc_length is 32 bits instead of 16)
-* to hold them all.
+* Eventually we should probably support the 12 byte READ DEFECT
+* DATA command. It supports a longer parameter list, which may be
+* necessary on newer drives with lots of defects. According to
+* the SBC-3 spec, drives are supposed to return an illegal request
+* if they have more defect data than will fit in 64K.
*/
- defect_list = malloc(dlist_length);
+ defect_list = malloc(max_dlist_length);
if (defect_list == NULL) {
warnx("can't malloc memory for defect list");
error = 1;
goto defect_bailout;
}
+ /*
+* We start off asking for just the header to determine how much
+* defect data is available. Some Hitachi drives return an error
+* if you ask for more data than the drive has. Once we know the
+* length, we retry the command with the returned length.
+*/
+ dlist_length = sizeof(struct scsi_read_defect_data_hdr_10);
+
rdd_cdb =(struct scsi_read_defect_data_10 *)&ccb->csio.cdb_io.cdb_bytes;
+retry:
+
+ lists_specified = 0;
+
/*
* cam_getccb() zeros the CCB header only. So we need to zero the
* payload portion of the ccb.
@@ -1917,6 +1931,51 @@ readdefects(struct cam_device *device, i
returned_length = scsi_2btoul(((struct
scsi_read_defect_data_hdr_10 *)defect_list)->length);
+ if (get_length != 0) {
+ get_length = 0;
+
+ if ((ccb->ccb_h.status & CAM_STATUS_MASK) ==
+CAM_SCSI_STATUS_ERROR) {
+ struct scsi_sense_data *sense;
+ int error_code, sense_key, asc, ascq;
+
+ sense = &ccb->csio.sense_data;
+
Re: svn commit: r237448 - head/lib/libedit
On Fri, Jun 22, 2012 at 06:01:23PM +, Pedro F. Giffuni wrote: > Author: pfg > Date: Fri Jun 22 18:01:22 2012 > New Revision: 237448 > URL: http://svn.freebsd.org/changeset/base/237448 > > Log: > Merge changes from upstream libedit. > > Our libedit has been diverging from the mainstream version > maintained in NetBSD. As a consequence it has been difficult > to do an appropriate MFV and we have been bringing only > partial updates. > > Here we update most of the files to at least match the > version available in NetBSD's snapshot of 20091228. This > version was chosen because it still doesn't include wide > character support (UTF-8), which involves many changes and > new files. > > From NetBSD's logs: > > Dec 15 22:13:33 2006 - editline.3 el.c el.h histedit.h > add EL_GETFP, and EL_SETFP. > > Apr 5 15:53:28 2008 - editline.3 el.c histedit.h readline.c > add EL_REFRESH for the benefit of readline > > Sep 10 15:45:37 2008 - common.c el.c read.c refresh.c sig.c term.c term.h > tty.c > Allow a single process to control multiple ttys (for pthreads using > _REENTRANT) > using multiple EditLine objects. > > Jan 18 12:17:24 2009 - el.c read.c readline.c > fix -Wsign-compare issues > > Feb 6 14:40:32 2009 - history.c > Plug memory leak, from MySQL. > > Feb 5 19:15:44 2009 - histedit.h read.c > match documentation in el_push > > Feb 6 13:14:37 2009 - vi.c > Portability fix. > > Feb 12 13:39:49 2009 - readline.c term.c > More fixes for existing portability stuff. > > Feb 15 21:24:13 2009 - el.h read.c > don't restart on EINTR, instead return NULL immediately. From Anon Ymous > > Feb 15 21:25:01 2009 - sig.c sig.h > in order for read() to return EINTR we need to use sigaction, not signal, > otherwise SA_RESTART is set. > > Feb 15 21:55:23 2009 - chared.c chared.h common.c emacs.c filecomplete.c > filecomplete.h key.c key.h read.c readline.c refresh.c search.c > term.c tokenizer.c tty.c vi.c > pass lint on _LP64. > > Feb 17 21:34:26 2009 - el.c histedit.h prompt.c prompt.h > allow for a prompt argument. > > Feb 18 15:04:40 2009 - sig.c > SA_RESTART for all signals but SIGINT. From Anon Ymous. > > Feb 19 15:20:22 2009 - read.c sig.c sig.h > reset and redraw on sigcont. From Anon Ymous. > > Feb 21 23:31:56 2009 - key.c key.h readline.c vi.c > more size_t stuff. > > Mar 10 20:46:15 2009 - editline.3 read.c > make el_gets set the count to -1 on error to distinguish between EOF and > error. > > Mar 31 17:38:27 2009 - editline.3 el.c histedit.h prompt.c prompt.h > refresh.c term.c term.h > Implement literal prompt sequences. Now someone can implement > RL_PROMPT_START_LITERAL/RL_PROMPT_END_LITERAL :-) > > Mar 31 21:33:17 2009 - term.c > cast to size_t to avoid sign / unsigned comparison warning. > > Apr 23 02:03 2009 - term.c > Apply patch (requested by msaitoh in ticket #2007): > Coverity CID 1668: Plug memory leak when malloc() failed.:55 2009 > > May 11 18:33:30 2009 - editline.3 el.c histedit.h > restore binary compatibility by providing new prompt functions that take > an extra literal character. > > May 19 21:45:14 2009 - refresh.c > always scroll when we advance past bottom. From Caleb Welton > cwelton at greenplum dot com. > > Jul 17 12:27:57 2009 - term.c > - off by one in the term.h case. > - make code more similar to tcsh (if we want to handle wide chars, this is > needed; for now it is a no-op) > > Jul 22 15:56:29 2009 - el.c > Move filename to the scope it is being used. > From Michael Cook mcook at bbn dot com > > Jul 22 15:57:00 2009 - read.c > Always initialize nread since it is an out param. > From Michael Cook mcook at bbn dot com > > Jul 22 18:25:26 2009 - el.c > Only need path if we have issetugid... From Anon Ymous > > Jul 25 21:19:23 2009 - el.c > Ignore comment lines in .editrc from Jess Thrysoee > > Sep 7 21:24:33 2009 > histedit.h history.c readline.c > apply apple patches from: > http://opensource.apple.com/source/libedit/libedit-11/patches/ > > Dec 28 21:52:43 2009 - refresh.c > Fix bug where tab completion on the second or > line that caused listing > ended up corrupting the display by an extra space in the beginning. Reported > by Mac Chan. > > Dec 28 22:15:36 2009 - refresh.c term.c > reduce diff with tcsh > > Obtained from: NetBSD > Tested by: bapt, jilles and current@ > MFC after: 1 week > > Modified: > head/lib/libedit/common.c > head/lib/libedit/editline.3 > head/lib/libedit/editrc.5 > head/lib/libedit/el.c > head/lib/libedit/el.h > head/lib/libedit/histedit.h > head/lib/libedit/history.c > head/lib/libedit/key.c > head/lib/libedit/key.h > head/lib/libedit/prompt.c > head/lib/libedit/prompt.h > head/lib/libedit/read.c > head/lib/libedit/refresh.c > head/lib/libedit/sear
Re: svn commit: r237448 - head/lib/libedit
-BEGIN PGP SIGNED MESSAGE- Hash: SHA256 As a general rule, please trim out replies. This list is worse than most in the sense that people often reply to hundreds of lines only to add a line or 2 at the bottom. By trimming your reply you both save electrons, and make it easier for people to see what you have to say. Doug -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.19 (FreeBSD) iQEcBAEBCAAGBQJP5NWTAAoJEFzGhvEaGryEpVsH/3I60AiJ6JODDA9mx+ufW9LG SAr/p13YGlswZxL5LOLCLOvwjNrvK6CJieb42XI+70PYkGAGH8syKunTy9s4EPuQ 2wXWHw4jlJaUfMTOFsXOvoZIFTrH0YPRt3qZic4TdfuxWlEkqrk4G37+lVNMhdsW 0MLJD/eJC5rV3Ve/YMKB2OSPK47mukOmY0spFXG/N2NQRryOuV8N3D6ycbWGDn7q YJwpO98YxbbyxLEV394rGwVCK8T6taJXRBOLMu9VFR0L7321TPuayPc3cLg+0hQC owr6vN2ga0uD7vjhoKbApBTTHzWAGYVWN1LU+rpE0/b5QGZByq2Muiz5SARBncI= =cYw2 -END PGP SIGNATURE- ___ svn-src-head@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r237458 - head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs
Author: mm
Date: Fri Jun 22 20:42:11 2012
New Revision: 237458
URL: http://svn.freebsd.org/changeset/base/237458
Log:
Import Illumos revision 13736:9f1d48e1681f
2901 ZFS receive fails for exabyte sparse files
References:
https://www.illumos.org/issues/2901
Obtained from:illumos (issue #2901)
MFC after:1 week
Modified:
head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_send.c
Modified: head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_send.c
==
--- head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_send.c Fri Jun
22 20:38:08 2012(r237457)
+++ head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_send.c Fri Jun
22 20:42:11 2012(r237458)
@@ -93,6 +93,9 @@ dump_free(dmu_sendarg_t *dsp, uint64_t o
{
struct drr_free *drrf = &(dsp->dsa_drr->drr_u.drr_free);
+ if (length != -1ULL && offset + length < offset)
+ length = -1ULL;
+
/*
* If there is a pending op, but it's not PENDING_FREE, push it out,
* since free block aggregation can only be done for blocks of the
___
svn-src-head@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r237459 - head/sys/netinet6
Author: bz
Date: Fri Jun 22 21:26:35 2012
New Revision: 237459
URL: http://svn.freebsd.org/changeset/base/237459
Log:
Just add a comment to further investigate when being closer to that code
again next time. The condition of the 2nd if() is very unlikely ever met.
Modified:
head/sys/netinet6/in6_src.c
Modified: head/sys/netinet6/in6_src.c
==
--- head/sys/netinet6/in6_src.c Fri Jun 22 20:42:11 2012(r237458)
+++ head/sys/netinet6/in6_src.c Fri Jun 22 21:26:35 2012(r237459)
@@ -597,6 +597,7 @@ selectroute(struct sockaddr_in6 *dstsock
if (ron->ro_rt == NULL) {
in6_rtalloc(ron, fibnum); /* multi path case? */
if (ron->ro_rt == NULL) {
+ /* XXX-BZ WT.? */
if (ron->ro_rt) {
RTFREE(ron->ro_rt);
ron->ro_rt = NULL;
___
svn-src-head@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r237460 - head/sys/dev/twa
Author: mav
Date: Fri Jun 22 21:46:41 2012
New Revision: 237460
URL: http://svn.freebsd.org/changeset/base/237460
Log:
Return CAM_SEL_TIMEOUT and CAM_DEV_NOT_THERE instead of CAM_TID_INVALID
and CAM_LUN_INVALID for case of missing devices. In removes tons of error
messages from CAM during bus scans.
Reported and tested by: Mike Tancsa
MFC after:3 days
Modified:
head/sys/dev/twa/tw_osl_cam.c
Modified: head/sys/dev/twa/tw_osl_cam.c
==
--- head/sys/dev/twa/tw_osl_cam.c Fri Jun 22 21:26:35 2012
(r237459)
+++ head/sys/dev/twa/tw_osl_cam.c Fri Jun 22 21:46:41 2012
(r237460)
@@ -678,9 +678,9 @@ tw_osl_complete_io(struct tw_cl_req_hand
ccb->ccb_h.status = CAM_REQ_CMP;
else {
if (req_pkt->status & TW_CL_ERR_REQ_INVALID_TARGET)
- ccb->ccb_h.status |= CAM_TID_INVALID;
+ ccb->ccb_h.status |= CAM_SEL_TIMEOUT;
else if (req_pkt->status & TW_CL_ERR_REQ_INVALID_LUN)
- ccb->ccb_h.status |= CAM_LUN_INVALID;
+ ccb->ccb_h.status |= CAM_DEV_NOT_THERE;
else if (req_pkt->status & TW_CL_ERR_REQ_SCSI_ERROR)
ccb->ccb_h.status |= CAM_SCSI_STATUS_ERROR;
else if (req_pkt->status & TW_CL_ERR_REQ_BUS_RESET)
___
svn-src-head@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r237463 - head/sys/dev/cxgbe
Author: np
Date: Fri Jun 22 22:59:42 2012
New Revision: 237463
URL: http://svn.freebsd.org/changeset/base/237463
Log:
Do not allocate extra vectors when adapter is not TOE
capable (or toecaps have been disallowed by the user).
+ one very minor unrelated cleanup in t4_sge.c
Modified:
head/sys/dev/cxgbe/t4_main.c
head/sys/dev/cxgbe/t4_sge.c
Modified: head/sys/dev/cxgbe/t4_main.c
==
--- head/sys/dev/cxgbe/t4_main.cFri Jun 22 22:50:50 2012
(r237462)
+++ head/sys/dev/cxgbe/t4_main.cFri Jun 22 22:59:42 2012
(r237463)
@@ -1308,10 +1308,12 @@ cfg_itype_and_nqueues(struct adapter *sc
iaq->nrxq10g = nrxq10g = t4_nrxq10g;
iaq->nrxq1g = nrxq1g = t4_nrxq1g;
#ifdef TCP_OFFLOAD
- iaq->nofldtxq10g = t4_nofldtxq10g;
- iaq->nofldtxq1g = t4_nofldtxq1g;
- iaq->nofldrxq10g = nofldrxq10g = t4_nofldrxq10g;
- iaq->nofldrxq1g = nofldrxq1g = t4_nofldrxq1g;
+ if (is_offload(sc)) {
+ iaq->nofldtxq10g = t4_nofldtxq10g;
+ iaq->nofldtxq1g = t4_nofldtxq1g;
+ iaq->nofldrxq10g = nofldrxq10g = t4_nofldrxq10g;
+ iaq->nofldrxq1g = nofldrxq1g = t4_nofldrxq1g;
+ }
#endif
for (itype = INTR_MSIX; itype; itype >>= 1) {
@@ -1380,7 +1382,8 @@ restart:
}
iaq->nrxq10g = min(n, nrxq10g);
#ifdef TCP_OFFLOAD
- iaq->nofldrxq10g = min(n, nofldrxq10g);
+ if (is_offload(sc))
+ iaq->nofldrxq10g = min(n, nofldrxq10g);
#endif
}
@@ -1395,7 +1398,8 @@ restart:
}
iaq->nrxq1g = min(n, nrxq1g);
#ifdef TCP_OFFLOAD
- iaq->nofldrxq1g = min(n, nofldrxq1g);
+ if (is_offload(sc))
+ iaq->nofldrxq1g = min(n, nofldrxq1g);
#endif
}
@@ -1408,7 +1412,8 @@ restart:
*/
iaq->nirq = iaq->nrxq10g = iaq->nrxq1g = 1;
#ifdef TCP_OFFLOAD
- iaq->nofldrxq10g = iaq->nofldrxq1g = 1;
+ if (is_offload(sc))
+ iaq->nofldrxq10g = iaq->nofldrxq1g = 1;
#endif
allocate:
Modified: head/sys/dev/cxgbe/t4_sge.c
==
--- head/sys/dev/cxgbe/t4_sge.c Fri Jun 22 22:50:50 2012(r237462)
+++ head/sys/dev/cxgbe/t4_sge.c Fri Jun 22 22:59:42 2012(r237463)
@@ -1031,7 +1031,7 @@ get_fl_payload(struct adapter *sc, struc
static int
t4_eth_rx(struct sge_iq *iq, const struct rss_header *rss, struct mbuf *m0)
{
- struct sge_rxq *rxq = (void *)iq;
+ struct sge_rxq *rxq = iq_to_rxq(iq);
struct ifnet *ifp = rxq->ifp;
const struct cpl_rx_pkt *cpl = (const void *)(rss + 1);
#ifdef INET
___
svn-src-head@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
Re: svn commit: r237350 - head/sys/dev/puc
On Jun 20, 2012, at 8:10 PM, Max Khon wrote:
>
> Modified: head/sys/dev/puc/puc_cfg.h
> ==
> --- head/sys/dev/puc/puc_cfg.hThu Jun 21 02:48:49 2012
> (r237349)
> +++ head/sys/dev/puc/puc_cfg.hThu Jun 21 03:10:48 2012
> (r237350)
> @@ -79,7 +79,7 @@ struct puc_cfg {
> int8_t ports;
> int8_t rid;/* Rid of first port */
> int8_t d_rid; /* Delta rid of next ports */
> - int8_t d_ofs; /* Delta offset of next ports */
> + int16_t d_ofs; /* Delta offset of next ports */
> puc_config_f*config_function;
> };
The purpose of the uint8_t is to minimize wastage. For the uncommon
cases there's the config_function. Given that uint8_t has been adequate
for a very long time, I think it's fair to say that the MOXA devices
have an uncommon configuration and that you shouldn't pessimize the
common case by extending the type of d_ofs and as a side-effect add
padding before it *and* after it -- the structure was optimal for both
ILP32 and LP64.
Please revert the above change and handle the uncommonly large offset
by extending the existing moxa configuration file. It's the precedence
that should have been followed anyway.
Thanks,
--
Marcel Moolenaar
mar...@xcllnt.net
___
svn-src-head@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
