Re: svn commit: r358733 - head/sys/sys
On 3/11/20 9:16 PM, Mateusz Guzik wrote:
On 3/10/20, Konstantin Belousov wrote:
On Tue, Mar 10, 2020 at 12:22:09AM +0100, Mateusz Guzik wrote:
On 3/9/20, Konstantin Belousov wrote:
On Mon, Mar 09, 2020 at 01:56:17AM +0100, Mateusz Guzik wrote:
On 3/8/20, Mateusz Guzik wrote:
Author: mjg
Date: Sun Mar 8 00:22:32 2020
New Revision: 358733
URL: https://svnweb.freebsd.org/changeset/base/358733
Log:
seqc: tidy up
- avoid hand-rolled read
- match begin/end in terms of fence style
There were off lists questions about this so let me clarify.
The first bit is a cosmetic change, but the second one is not.
Modified:
head/sys/sys/seqc.h
Modified: head/sys/sys/seqc.h
==
--- head/sys/sys/seqc.h Sat Mar 7 15:37:23 2020(r358732)
+++ head/sys/sys/seqc.h Sun Mar 8 00:22:32 2020(r358733)
@@ -66,7 +66,8 @@ static __inline void
seqc_write_end(seqc_t *seqcp)
{
- atomic_store_rel_int(seqcp, *seqcp + 1);
+ atomic_thread_fence_rel();
+ *seqcp += 1;
MPASS(!seqc_in_modify(*seqcp));
critical_exit();
}
For correct operation the counter has to be modified *before* any work
is done and *after* it is completed.
Consider a trivial case:
seqc++;
foo[0] = 0;
foo[1] = 1;
seqc++;
There are 2 ways in which this can be mucked with:
- the compiler can be looking at reordering or reworking the increment
(e.g., to collapse it to just += 2 instead). a compiler barrier
prevents that.
- even with generated machine code which increments in correct places
the cpu can be looking at reordering the operation (which is heavily
dependent on architecture), in particular it could end up issuing
seqc++; foo[1] = 1; which would defeat the point. This is where
release fences come in.
With the patched code there is:
seqc++;
atomic_thread_fence_rel(); /* make sure seqc++ is issued before foo
starts getting modified */
foo[0] = 0;
foo[1] = 1;
atomic_thread_fence_rel(); /* make sure modifications to foo don't
leak past this spot */
seqc++;
In comparison, the previous code was:
seqc++;
atomic_thread_fence_rel(); /* make sure seqc++ is issued before foo
starts getting modified */
foo[0] = 0;
foo[1] = 1;
atomic_store_rel_int(seqcp, *seqcp + 1); /* make sure modifications to
too don't leak past this spot and update seqc immediately */
There are 2 differences here -- one is an improvement and second one
does not matter:
the win is: the previous code forces the compiler to compute an
incremented value and then store it. On amd64 this translates to the
following (with rdx holding the address of the counter):
mov(%rdx),%eax /* load the value */
add$0x1,%eax/* add 1 */
mov%eax,(%rdx) /* store it */
On patched kernel this is:
addl $0x1,(%rdx)
which is clearly much nicer.
I am not sure that the new code on amd64 is much nicer.
But the point was that on non-amd64, i.e. armv8 and potentially on
powerpc 3.0, the patch substitutes release store with full barrier. The
later is (much) slower.
If an arch performs something significantly more expensive here it's
probably a performance bug in the port.
It is a question of how much hardware support for fine-grained fences.
Relaxed architectures add some optimized operations already, but did not
(yet) implemented complete C11 optimized set.
I think they eventually converge.
Until they did not, I think it is more important to not pessimize
to-be Tier1 arches than to collapse three fast integer instructions
on amd64 into one rw op.
But perhaps more importantly
there are significantly more frequent users of
atomic_thread_fence_rel, like refcount(9). If the issue is real that
should be looked at.
For refcount(9) use of release fence removal, we would need
atomic_fetchadd_rel_int().
To reiterate, if atomic_thread_fence_rel is indeed very expensive on
certain architectures compared to other _rel variants, the kernel is
already heavily pessimized because of use of said standalone fence in
refcount(9) and few other places and this needs to be fixed.
Just to be clear, this is not just a matter of whether our release
fences are poorly or inefficiently implemented. Release fences (as
defined by the C/C++ standards that inspire our atomics) have
*different* semantics than release stores. Specifically, a release fence
places more ordering restrictions on the nearby store accesses than a
release store does. (In fact, "man 9 atomic" describes this.) Thus, it
is almost inevitable that a release fence will have a higher cost than
the best possible implementation of a release store.
If the cost can't be reduced (e.g., perhaps this can do a no-op store
+ fence somewhere akin to how lock xadd 0,(%esp) trickery used to
work), then the atomic APIs do indeed not to be completed w.r.t.
adding all the missing acq/rel variants (that's at least fetchadd and
swap).
We can also extend the API to do relevant addition/subtraction without
gua
Re: svn commit: r358733 - head/sys/sys
On 12/03/2020 09:48, Alan Cox wrote: > Just to be clear, this is not just a matter of whether our release fences are > poorly or inefficiently implemented. Release fences (as defined by the C/C++ > standards that inspire our atomics) have *different* semantics than release > stores. Specifically, a release fence places more ordering restrictions on the > nearby store accesses than a release store does. (In fact, "man 9 atomic" > describes this.) Thus, it is almost inevitable that a release fence will > have a > higher cost than the best possible implementation of a release store. For people like me who are new to these things here is a good explanation (perhaps a bad title, though): https://preshing.com/20131125/acquire-and-release-fences-dont-work-the-way-youd-expect/ -- Andriy Gapon ___ svn-src-head@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r358907 - head
Author: dim
Date: Thu Mar 12 11:39:04 2020
New Revision: 358907
URL: https://svnweb.freebsd.org/changeset/base/358907
Log:
Allow -DNO_CLEAN build across r358851.
The openmp 10.0.0 import renamed one .c file to .cpp, and this is
something our dependency system does not handle correctly. Add another
ad-hoc cleanup to get rid of the stale dependency.
PR: 244251
MFC after:6 weeks
X-MFC-With: 358851
Modified:
head/Makefile.inc1
Modified: head/Makefile.inc1
==
--- head/Makefile.inc1 Thu Mar 12 06:45:08 2020(r358906)
+++ head/Makefile.inc1 Thu Mar 12 11:39:04 2020(r358907)
@@ -924,6 +924,15 @@ _sanity_check: .PHONY .MAKE
_cleanobj_fast_depend_hack: .PHONY
# Syscall stubs rewritten in C and obsolete MD assembly implementations
# Date SVN Rev Syscalls/Changes
+# 20200310 r358851 rename of openmp's ittnotify_static.c to .cpp
+.for f in ittnotify_static
+ @if [ -e "${OBJTOP}/lib/libomp/.depend.${f}.pico" ] && \
+ egrep -qw '${f}\.c' ${OBJTOP}/lib/libomp/.depend.${f}.pico; then \
+ echo "Removing stale dependencies for ${f}"; \
+ rm -f ${OBJTOP}/lib/libomp/.depend.${f}.* \
+ ${LIBCOMPAT:D${LIBCOMPAT_OBJTOP}/lib/libomp/.depend.${f}.*};
\
+ fi
+.endfor
# 20191009 r353340 removal of opensolaris_atomic.S (also r353381)
.if ${MACHINE} != i386
.for f in opensolaris_atomic
___
svn-src-head@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r358908 - in head/sys: conf modules powerpc/conf
Author: luporl
Date: Thu Mar 12 12:47:10 2020
New Revision: 358908
URL: https://svnweb.freebsd.org/changeset/base/358908
Log:
Enable ixl device on PowerPC64
The ixl driver now works on PowerPC64 and may be compiled in-kernel and
as a module.
Reviewed by: alfredo, erj
Sponsored by: Eldorado Research Institute (eldorado.org.br)
Differential Revision:https://reviews.freebsd.org/D23974
Modified:
head/sys/conf/files.powerpc
head/sys/conf/options.powerpc
head/sys/modules/Makefile
head/sys/powerpc/conf/GENERIC64
Modified: head/sys/conf/files.powerpc
==
--- head/sys/conf/files.powerpc Thu Mar 12 11:39:04 2020(r358907)
+++ head/sys/conf/files.powerpc Thu Mar 12 12:47:10 2020(r358908)
@@ -38,6 +38,43 @@ dev/iicbus/max6690.c optionalmax6690 powermac
dev/iicbus/ofw_iicbus.coptionaliicbus aim
dev/ipmi/ipmi.coptionalipmi
dev/ipmi/ipmi_opal.c optionalpowernv ipmi
+dev/ixl/if_ixl.c optionalixl pci powerpc64 \
+ compile-with "${NORMAL_C} -I$S/dev/ixl"
+dev/ixl/ixl_pf_main.c optionalixl pci powerpc64 \
+ compile-with "${NORMAL_C} -I$S/dev/ixl"
+dev/ixl/ixl_pf_qmgr.c optionalixl pci powerpc64 \
+ compile-with "${NORMAL_C} -I$S/dev/ixl"
+dev/ixl/ixl_pf_iov.c optionalixl pci pci_iov powerpc64 \
+ compile-with "${NORMAL_C} -I$S/dev/ixl"
+dev/ixl/ixl_pf_i2c.c optionalixl pci powerpc64 \
+ compile-with "${NORMAL_C} -I$S/dev/ixl"
+dev/ixl/if_iavf.c optionaliavf pci powerpc64 \
+ compile-with "${NORMAL_C} -I$S/dev/ixl"
+dev/ixl/iavf_vc.c optionaliavf pci powerpc64 \
+ compile-with "${NORMAL_C} -I$S/dev/ixl"
+dev/ixl/ixl_txrx.c optionalixl pci powerpc64 | \
+ iavf pci powerpc64 \
+ compile-with "${NORMAL_C} -I$S/dev/ixl"
+dev/ixl/i40e_osdep.c optionalixl pci powerpc64 | \
+ iavf pci powerpc64 \
+ compile-with "${NORMAL_C} -I$S/dev/ixl"
+dev/ixl/i40e_lan_hmc.c optionalixl pci powerpc64 | \
+ iavf pci powerpc64 \
+ compile-with "${NORMAL_C} -I$S/dev/ixl"
+dev/ixl/i40e_hmc.c optionalixl pci powerpc64 | \
+ iavf pci powerpc64 \
+ compile-with "${NORMAL_C} -I$S/dev/ixl"
+dev/ixl/i40e_common.c optionalixl pci powerpc64 | \
+ iavf pci powerpc64 \
+ compile-with "${NORMAL_C} -I$S/dev/ixl"
+dev/ixl/i40e_nvm.c optionalixl pci powerpc64 | \
+ iavf pci powerpc64 \
+ compile-with "${NORMAL_C} -I$S/dev/ixl"
+dev/ixl/i40e_adminq.c optionalixl pci powerpc64 | \
+ iavf pci powerpc64 \
+ compile-with "${NORMAL_C} -I$S/dev/ixl"
+dev/ixl/i40e_dcb.c optionalixl pci powerpc64 \
+ compile-with "${NORMAL_C} -I$S/dev/ixl"
# Most ofw stuff below is brought in by conf/files for options FDT, but
# we always want it, even on non-FDT platforms.
dev/fdt/simplebus.cstandard
Modified: head/sys/conf/options.powerpc
==
--- head/sys/conf/options.powerpc Thu Mar 12 11:39:04 2020
(r358907)
+++ head/sys/conf/options.powerpc Thu Mar 12 12:47:10 2020
(r358908)
@@ -37,3 +37,5 @@ OFWCONS_POLL_HZ opt_ofw.h
# AGP debugging support
AGP_DEBUG opt_agp.h
+# iWARP client interface support in ixl
+IXL_IW opt_ixl.h
Modified: head/sys/modules/Makefile
==
--- head/sys/modules/Makefile Thu Mar 12 11:39:04 2020(r358907)
+++ head/sys/modules/Makefile Thu Mar 12 12:47:10 2020(r358908)
@@ -766,6 +766,7 @@ _virtio=virtio
.if ${MACHINE_ARCH} == "powerpc64"
_ipmi= ipmi
+_ixl= ixl
_nvram=opal_nvram
.endif
.if ${MACHINE_ARCH} == "powerpc64" || ${MACHINE_ARCH} == "powerpc"
Modified: head/sys/powerpc/conf/GENERIC64
==
--- head/sys/powerpc/conf/GENERIC64 Thu Mar 12 11:39:04 2020
(r358907)
+++ head/sys/powerpc/conf/GENERIC64 Thu Mar 12 12:47:10 2020
(r358908)
@@ -168,6 +168,7 @@ device iflib
# Ethernet hardware
device em # Intel PRO/1000 Gigabit Ethernet Family
device ix # Intel PRO/10GbE PCIE PF Ethernet Family
+device ixl
svn commit: r358909 - head
Author: emaste
Date: Thu Mar 12 13:42:08 2020
New Revision: 358909
URL: https://svnweb.freebsd.org/changeset/base/358909
Log:
Extend r358907 to explicitly remove stale lib32 dependency
After r325072 stale lib32 dependencies were not remooved. A more
holistic approach is needed to address this but for the immediate issue
(-DNO_CLEAN builds across r358851) just readd the explicit lib32 path.
Reported by: dim
Sponsored by: The FreeBSD Foundation
Modified:
head/Makefile.inc1
Modified: head/Makefile.inc1
==
--- head/Makefile.inc1 Thu Mar 12 12:47:10 2020(r358908)
+++ head/Makefile.inc1 Thu Mar 12 13:42:08 2020(r358909)
@@ -930,6 +930,7 @@ _cleanobj_fast_depend_hack: .PHONY
egrep -qw '${f}\.c' ${OBJTOP}/lib/libomp/.depend.${f}.pico; then \
echo "Removing stale dependencies for ${f}"; \
rm -f ${OBJTOP}/lib/libomp/.depend.${f}.* \
+ ${OBJTOP}/obj-lib32/lib/libomp/.depend.${f}.* \
${LIBCOMPAT:D${LIBCOMPAT_OBJTOP}/lib/libomp/.depend.${f}.*};
\
fi
.endfor
___
svn-src-head@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r358910 - head
Author: emaste
Date: Thu Mar 12 14:01:17 2020
New Revision: 358910
URL: https://svnweb.freebsd.org/changeset/base/358910
Log:
Makefile.inc1: move dependency hack comment to the block it applies to
Modified:
head/Makefile.inc1
Modified: head/Makefile.inc1
==
--- head/Makefile.inc1 Thu Mar 12 13:42:08 2020(r358909)
+++ head/Makefile.inc1 Thu Mar 12 14:01:17 2020(r358910)
@@ -922,7 +922,6 @@ _sanity_check: .PHONY .MAKE
# tree changes, particularly with respect to removing source files and
# replacing generated files. Handle these cases here in an ad-hoc fashion.
_cleanobj_fast_depend_hack: .PHONY
-# Syscall stubs rewritten in C and obsolete MD assembly implementations
# Date SVN Rev Syscalls/Changes
# 20200310 r358851 rename of openmp's ittnotify_static.c to .cpp
.for f in ittnotify_static
@@ -934,6 +933,7 @@ _cleanobj_fast_depend_hack: .PHONY
${LIBCOMPAT:D${LIBCOMPAT_OBJTOP}/lib/libomp/.depend.${f}.*};
\
fi
.endfor
+# Syscall stubs rewritten in C and obsolete MD assembly implementations
# 20191009 r353340 removal of opensolaris_atomic.S (also r353381)
.if ${MACHINE} != i386
.for f in opensolaris_atomic
___
svn-src-head@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r358911 - in head/sys/netinet: . tcp_stacks
Author: tuexen
Date: Thu Mar 12 15:37:41 2020
New Revision: 358911
URL: https://svnweb.freebsd.org/changeset/base/358911
Log:
Use KMOD_TCPSTAT_INC instead of TCPSTAT_INC for RACK and BBR, since
these are kernel modules. Also add a KMOD_TCPSTAT_ADD and use that
instead of TCPSTAT_ADD.
Reviewed by: jtl@, rrs@
MFC after:1 week
Sponsored by: Netflix, Inc.
Differential Revision:https://reviews.freebsd.org/D23904
Modified:
head/sys/netinet/tcp_input.c
head/sys/netinet/tcp_stacks/bbr.c
head/sys/netinet/tcp_stacks/rack.c
head/sys/netinet/tcp_stacks/rack_bbr_common.c
head/sys/netinet/tcp_var.h
Modified: head/sys/netinet/tcp_input.c
==
--- head/sys/netinet/tcp_input.cThu Mar 12 14:01:17 2020
(r358910)
+++ head/sys/netinet/tcp_input.cThu Mar 12 15:37:41 2020
(r358911)
@@ -266,14 +266,14 @@ VNET_SYSUNINIT(tcp_vnet_uninit, SI_SUB_PROTO_IFATTACHD
#endif /* VIMAGE */
/*
- * Kernel module interface for updating tcpstat. The argument is an index
+ * Kernel module interface for updating tcpstat. The first argument is an
index
* into tcpstat treated as an array.
*/
void
-kmod_tcpstat_inc(int statnum)
+kmod_tcpstat_add(int statnum, int val)
{
- counter_u64_add(VNET(tcpstat)[statnum], 1);
+ counter_u64_add(VNET(tcpstat)[statnum], val);
}
#ifdef TCP_HHOOK
Modified: head/sys/netinet/tcp_stacks/bbr.c
==
--- head/sys/netinet/tcp_stacks/bbr.c Thu Mar 12 14:01:17 2020
(r358910)
+++ head/sys/netinet/tcp_stacks/bbr.c Thu Mar 12 15:37:41 2020
(r358911)
@@ -1866,7 +1866,7 @@ bbr_progress_timeout_check(struct tcp_bbr *bbr)
bbr_log_progress_event(bbr, bbr->rc_tp, ticks,
PROGRESS_DROP, __LINE__);
BBR_STAT_INC(bbr_progress_drops);
#ifdef NETFLIX_STATS
- TCPSTAT_INC(tcps_progdrops);
+ KMOD_TCPSTAT_INC(tcps_progdrops);
#endif
return (1);
}
@@ -4067,7 +4067,7 @@ bbr_cong_signal(struct tcpcb *tp, struct tcphdr *th, u
}
break;
case CC_RTO_ERR:
- TCPSTAT_INC(tcps_sndrexmitbad);
+ KMOD_TCPSTAT_INC(tcps_sndrexmitbad);
/* RTO was unnecessary, so reset everything. */
bbr_reset_lt_bw_sampling(bbr, bbr->r_ctl.rc_rcvtime);
if (bbr->rc_bbr_state != BBR_STATE_PROBE_RTT) {
@@ -4808,7 +4808,7 @@ bbr_timeout_delack(struct tcpcb *tp, struct tcp_bbr *b
bbr_log_to_event(bbr, cts, BBR_TO_FRM_DELACK);
tp->t_flags &= ~TF_DELACK;
tp->t_flags |= TF_ACKNOW;
- TCPSTAT_INC(tcps_delack);
+ KMOD_TCPSTAT_INC(tcps_delack);
bbr->r_ctl.rc_hpts_flags &= ~PACE_TMR_DELACK;
return (0);
}
@@ -4840,7 +4840,7 @@ bbr_timeout_persist(struct tcpcb *tp, struct tcp_bbr *
*/
bbr_log_to_event(bbr, cts, BBR_TO_FRM_PERSIST);
bbr->r_ctl.rc_hpts_flags &= ~PACE_TMR_PERSIT;
- TCPSTAT_INC(tcps_persisttimeo);
+ KMOD_TCPSTAT_INC(tcps_persisttimeo);
/*
* Have we exceeded the user specified progress time?
*/
@@ -4857,7 +4857,7 @@ bbr_timeout_persist(struct tcpcb *tp, struct tcp_bbr *
if (tp->t_rxtshift == TCP_MAXRXTSHIFT &&
(ticks - tp->t_rcvtime >= tcp_maxpersistidle ||
ticks - tp->t_rcvtime >= TCP_REXMTVAL(tp) * tcp_totbackoff)) {
- TCPSTAT_INC(tcps_persistdrop);
+ KMOD_TCPSTAT_INC(tcps_persistdrop);
tcp_set_inp_to_drop(bbr->rc_inp, ETIMEDOUT);
goto out;
}
@@ -4873,7 +4873,7 @@ bbr_timeout_persist(struct tcpcb *tp, struct tcp_bbr *
*/
if (tp->t_state > TCPS_CLOSE_WAIT &&
(ticks - tp->t_rcvtime) >= TCPTV_PERSMAX) {
- TCPSTAT_INC(tcps_persistdrop);
+ KMOD_TCPSTAT_INC(tcps_persistdrop);
tcp_set_inp_to_drop(bbr->rc_inp, ETIMEDOUT);
goto out;
}
@@ -4916,7 +4916,7 @@ bbr_timeout_keepalive(struct tcpcb *tp, struct tcp_bbr
* Keep-alive timer went off; send something or drop connection if
* idle for too long.
*/
- TCPSTAT_INC(tcps_keeptimeo);
+ KMOD_TCPSTAT_INC(tcps_keeptimeo);
if (tp->t_state < TCPS_ESTABLISHED)
goto dropit;
if ((V_tcp_always_keepalive || inp->inp_socket->so_options &
SO_KEEPALIVE) &&
@@ -4933,7 +4933,7 @@ bbr_timeout_keepalive(struct tcpcb *tp, struct tcp_bbr
* protocol spec, this requires the correspondent TCP to
* respond.
*/
- TCPSTAT_INC(tcps_keepprobe);
+ KMOD_TCPSTAT_INC(tcps_keepprobe);
t_template = tcpip_maketemplate(inp);
if (t_template) {
svn commit: r358912 - head/tools/build/mk
Author: ngie Date: Thu Mar 12 16:06:26 2020 New Revision: 358912 URL: https://svnweb.freebsd.org/changeset/base/358912 Log: Remove /usr/share/snmp/defs/tc.def with delete-old if MK_BSNMP == no This removes a lingering file on new installs of CURRENT, originally added in r345797. MFC after:1 week MFC with: r345797 Modified: head/tools/build/mk/OptionalObsoleteFiles.inc Modified: head/tools/build/mk/OptionalObsoleteFiles.inc == --- head/tools/build/mk/OptionalObsoleteFiles.inc Thu Mar 12 15:37:41 2020(r358911) +++ head/tools/build/mk/OptionalObsoleteFiles.inc Thu Mar 12 16:06:26 2020(r358912) @@ -972,6 +972,7 @@ OLD_FILES+=usr/share/snmp/defs/mibII_tree.def OLD_FILES+=usr/share/snmp/defs/netgraph_tree.def OLD_FILES+=usr/share/snmp/defs/pf_tree.def OLD_FILES+=usr/share/snmp/defs/target_tree.def +OLD_FILES+=usr/share/snmp/defs/tc.def OLD_FILES+=usr/share/snmp/defs/tree.def OLD_FILES+=usr/share/snmp/defs/usm_tree.def OLD_FILES+=usr/share/snmp/defs/vacm_tree.def ___ svn-src-head@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
Re: svn commit: r358655 - head/sbin/mount_nfs
On Thu, Mar 05, 2020 at 11:15:04PM +0100, Dimitry Andric wrote: D> > S> > Why don't just declare the buffer as: D> > S> > D> > S> > struct if_msghdr buf; D> > S> > D> > S> > and then do: D> > S> > D> > S> > nread = read(s, &buf, sizeof buf); D> > S> > D> > S> > ? You are never reading more than one if_msghdr anyway, and then there D> > S> > is no need to cast anything. D> > S> D> > S> My inspiration: route socket can return other messages (man 4 route) D> > D> > Yes, exactly. We don't know what size next datagram is going to be. D> D> Oh, in that case this code seems completely wrong. How do you know the D> full datagram will be delivered with one read() call? D> D> If it always is, then there is no need to read more than the size of D> struct if_msghdr, since you are not using any data beyond it. So in D> that case, you can suffice with read(..., sizeof(if_msghdr)). D> D> If the read() call will return partial results, you must repeatedly call D> it in a loop, until you either reach EOF, or have enough data. In that D> case, a buffer with the size of if_msghdr is also enough, since you D> never need to read beyond that. Sorry for delayed answer. The routing socket is a datagram socket, it isn't like TCP, it can't deliver partial datagrams. If we don't supply enough space for datagram that arrived, it won't be delivered. So the right solution is suppling plenty of space, but parse only part we are interested in. -- Gleb Smirnoff ___ svn-src-head@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
Re: svn commit: r358902 - head/lib/libc/powerpc64/string
To be clear here, I meant non qword-aligned, as it was the VSX implementation that was actually tripping up here. The problem manifested when the source and destination didn't match on the least significant nybble. Apparently it's also somewhat uncommon for programs to ever do anything with the return value as well. On Wed, Mar 11, 2020, at 6:34 PM, Brandon Bergren wrote: > Author: bdragon > Date: Wed Mar 11 23:34:44 2020 > New Revision: 358902 > URL: https://svnweb.freebsd.org/changeset/base/358902 > > Log: > Fix r358688 -- Remember to actually save r3 before processing. > > Crash was noticed by pkubaj building gcc9. > > Apparently non dword-aligned char pointers are somewhat rare in the wild. > > Reported by:pkubaj > Sponsored by: Tag1 Consulting, Inc. > > Modified: > head/lib/libc/powerpc64/string/memcpy.S > > Modified: head/lib/libc/powerpc64/string/memcpy.S > == > --- head/lib/libc/powerpc64/string/memcpy.S Wed Mar 11 22:25:45 2020 > (r358901) > +++ head/lib/libc/powerpc64/string/memcpy.S Wed Mar 11 23:34:44 2020 > (r358902) > @@ -58,8 +58,8 @@ ENTRY(FN_NAME) > andi. %r7, %r4, ALIGN_MASK > cmpd%r8, %r7 > mr %r7, %r5 > - bne .Lcopy_remaining_fix_index_byte > mr %r8, %r3/* save dst */ > + bne .Lcopy_remaining_fix_index_byte > > /* align src */ > .Lalignment_loop: > -- Brandon Bergren bdra...@imap.cc ___ svn-src-head@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r358916 - head/lib/libc/gen
Author: 0mp (doc,ports committer)
Date: Thu Mar 12 18:28:23 2020
New Revision: 358916
URL: https://svnweb.freebsd.org/changeset/base/358916
Log:
ftw.3: Follow style(9) in the example
Reported by: oshogbo
Approved by: bcr (mentor)
MFC after:2 weeks
Differential Revision:https://reviews.freebsd.org/D24043
Modified:
head/lib/libc/gen/ftw.3
Modified: head/lib/libc/gen/ftw.3
==
--- head/lib/libc/gen/ftw.3 Thu Mar 12 16:15:03 2020(r358915)
+++ head/lib/libc/gen/ftw.3 Thu Mar 12 18:28:23 2020(r358916)
@@ -20,7 +20,7 @@
.\"
.\" $FreeBSD$
.\"
-.Dd March 5, 2020
+.Dd March 12, 2020
.Dt FTW 3
.Os
.Sh NAME
@@ -209,6 +209,7 @@ nftw_callback(const char *path, const struct stat *sb,
int
main(int argc, char **argv)
{
+
if (argc != 2) {
printf("Usage %s \\n", argv[0]);
return (EX_USAGE);
___
svn-src-head@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r358919 - head/usr.bin/logger
Author: oshogbo
Date: Thu Mar 12 19:55:51 2020
New Revision: 358919
URL: https://svnweb.freebsd.org/changeset/base/358919
Log:
logger: capsicumize
Submitted by: Tiger Gao
Sponsored by: The FreeBSD Foundation
Differential Revision:https://reviews.freebsd.org/D23744
Modified:
head/usr.bin/logger/Makefile
head/usr.bin/logger/logger.c
Modified: head/usr.bin/logger/Makefile
==
--- head/usr.bin/logger/MakefileThu Mar 12 19:10:53 2020
(r358918)
+++ head/usr.bin/logger/MakefileThu Mar 12 19:55:51 2020
(r358919)
@@ -9,4 +9,10 @@ PROG= logger
CFLAGS+= -DINET6
.endif
+.if ${MK_CASPER} != "no"
+LIBADD+=casper
+LIBADD+=cap_syslog
+CFLAGS+=-DWITH_CASPER
+.endif
+
.include
Modified: head/usr.bin/logger/logger.c
==
--- head/usr.bin/logger/logger.cThu Mar 12 19:10:53 2020
(r358918)
+++ head/usr.bin/logger/logger.cThu Mar 12 19:55:51 2020
(r358919)
@@ -44,10 +44,12 @@ static char sccsid[] = "@(#)logger.c8.1 (Berkeley)
6/
#include
__FBSDID("$FreeBSD$");
+#include
#include
#include
#include
+#include
#include
#include
#include
@@ -57,6 +59,9 @@ __FBSDID("$FreeBSD$");
#include
#include
+#include
+#include
+
#defineSYSLOG_NAMES
#include
@@ -76,6 +81,7 @@ static void logmessage(int, const char *, const char *
struct socks *, ssize_t, const char *);
static voidusage(void);
+static cap_channel_t *capsyslog;
#ifdef INET6
static int family = PF_UNSPEC; /* protocol family (IPv4, IPv6 or both) */
#else
@@ -92,12 +98,13 @@ static int send_to_all = 0; /* send message to all IPv
int
main(int argc, char *argv[])
{
+ cap_channel_t *capcas;
struct socks *socks;
ssize_t nsock;
time_t now;
int ch, logflags, pri;
char *tag, *host, buf[1024], *timestamp, tbuf[26],
- *hostname, hbuf[MAXHOSTNAMELEN];
+ *hostname, hbuf[MAXHOSTNAMELEN], *pristr;
const char *svcname, *src;
tag = NULL;
@@ -107,6 +114,7 @@ main(int argc, char *argv[])
src = NULL;
socks = NULL;
pri = LOG_USER | LOG_NOTICE;
+ pristr = NULL;
logflags = 0;
unsetenv("TZ");
while ((ch = getopt(argc, argv, "46Af:H:h:iP:p:S:st:")) != -1)
@@ -140,7 +148,7 @@ main(int argc, char *argv[])
svcname = optarg;
break;
case 'p': /* priority */
- pri = pencode(optarg);
+ pristr = optarg;
break;
case 's': /* log to standard error */
logflags |= LOG_PERROR;
@@ -168,12 +176,25 @@ main(int argc, char *argv[])
nsock = 0;
}
+ capcas = cap_init();
+ if (capcas == NULL)
+ err(1, "Unable to contact Casper");
+ caph_cache_catpages();
+ caph_cache_tzdata();
+ if (caph_enter() < 0)
+ err(1, "Unable to enter capability mode");
+ capsyslog = cap_service_open(capcas, "system.syslog");
+ if (capsyslog == NULL)
+ err(1, "Unable to open system.syslog service");
+ cap_close(capcas);
+
+ if (pristr != NULL)
+ pri = pencode(pristr);
if (tag == NULL)
tag = getlogin();
/* setup for logging */
if (host == NULL)
- openlog(tag, logflags, 0);
- (void) fclose(stdout);
+ cap_openlog(capsyslog, tag, logflags, 0);
(void )time(&now);
(void )ctime_r(&now, tbuf);
@@ -349,7 +370,7 @@ logmessage(int pri, const char *timestamp, const char
int len, i, lsent;
if (nsock == 0) {
- syslog(pri, "%s", buf);
+ cap_syslog(capsyslog, pri, "%s", buf);
return;
}
if ((len = asprintf(&line, "<%d>%s %s %s: %s", pri, timestamp,
___
svn-src-head@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r358922 - head/sys/kern
Author: cem
Date: Thu Mar 12 21:26:36 2020
New Revision: 358922
URL: https://svnweb.freebsd.org/changeset/base/358922
Log:
kern_shutdown: Add missing EKCD ifdef
Submitted by: Puneeth Jothaiah
Reviewed by: bdrewery
Sponsored by: Dell EMC Isilon
Modified:
head/sys/kern/kern_shutdown.c
Modified: head/sys/kern/kern_shutdown.c
==
--- head/sys/kern/kern_shutdown.c Thu Mar 12 20:59:00 2020
(r358921)
+++ head/sys/kern/kern_shutdown.c Thu Mar 12 21:26:36 2020
(r358922)
@@ -1231,6 +1231,7 @@ dumper_insert(const struct dumperinfo *di_template, co
#endif
}
if (kda->kda_compression != KERNELDUMP_COMP_NONE) {
+#ifdef EKCD
/*
* We can't support simultaneous unpadded block cipher
* encryption and compression because there is no guarantee the
@@ -1241,6 +1242,7 @@ dumper_insert(const struct dumperinfo *di_template, co
error = EOPNOTSUPP;
goto cleanup;
}
+#endif
newdi->kdcomp = kerneldumpcomp_create(newdi,
kda->kda_compression);
if (newdi->kdcomp == NULL) {
___
svn-src-head@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r358923 - in head: lib/libpmcstat usr.sbin/pmcstat
Author: freqlabs
Date: Thu Mar 12 23:04:40 2020
New Revision: 358923
URL: https://svnweb.freebsd.org/changeset/base/358923
Log:
libpmcstat: Try /boot/modules if module not found
Modules from ports/pkg are commonly installed to /boot/modules rather than to
the same directory the kernel resides in. Look there if a module is not found
next to the kernel.
Submitted by: mmacy
Reported by: Nick Principe
Approved by: mmacy (mentor)
MFC after:2 weeks
Sponsored by: iXsystems, Inc.
Modified:
head/lib/libpmcstat/libpmcstat_image.c
head/usr.sbin/pmcstat/pmcstat.8
Modified: head/lib/libpmcstat/libpmcstat_image.c
==
--- head/lib/libpmcstat/libpmcstat_image.c Thu Mar 12 21:26:36 2020
(r358922)
+++ head/lib/libpmcstat/libpmcstat_image.c Thu Mar 12 23:04:40 2020
(r358923)
@@ -278,6 +278,7 @@ pmcstat_image_get_elf_params(struct pmcstat_image *ima
GElf_Shdr sh;
enum pmcstat_image_type image_type;
char buffer[PATH_MAX];
+ char buffer_modules[PATH_MAX];
assert(image->pi_type == PMCSTAT_IMAGE_UNKNOWN);
@@ -292,23 +293,32 @@ pmcstat_image_get_elf_params(struct pmcstat_image *ima
assert(path != NULL);
/*
-* Look for kernel modules under FSROOT/KERNELPATH/NAME,
-* and user mode executable objects under FSROOT/PATHNAME.
+* Look for kernel modules under FSROOT/KERNELPATH/NAME and
+* FSROOT/boot/modules/NAME, and user mode executable objects
+* under FSROOT/PATHNAME.
*/
- if (image->pi_iskernelmodule)
+ if (image->pi_iskernelmodule) {
(void) snprintf(buffer, sizeof(buffer), "%s%s/%s",
args->pa_fsroot, args->pa_kernel, path);
- else
+ (void) snprintf(buffer_modules, sizeof(buffer_modules),
+ "%s/boot/modules/%s", args->pa_fsroot, path);
+ } else {
(void) snprintf(buffer, sizeof(buffer), "%s%s",
args->pa_fsroot, path);
+ }
e = NULL;
- if ((fd = open(buffer, O_RDONLY, 0)) < 0) {
+ fd = open(buffer, O_RDONLY, 0);
+ if (fd < 0 && !image->pi_iskernelmodule) {
warnx("WARNING: Cannot open \"%s\".",
buffer);
goto done;
}
-
+ if (fd < 0 && (fd = open(buffer_modules, O_RDONLY, 0)) < 0) {
+ warnx("WARNING: Cannot open \"%s\" or \"%s\".",
+ buffer, buffer_modules);
+ goto done;
+ }
if (elf_version(EV_CURRENT) == EV_NONE) {
warnx("WARNING: failed to init elf\n");
goto done;
Modified: head/usr.sbin/pmcstat/pmcstat.8
==
--- head/usr.sbin/pmcstat/pmcstat.8 Thu Mar 12 21:26:36 2020
(r358922)
+++ head/usr.sbin/pmcstat/pmcstat.8 Thu Mar 12 23:04:40 2020
(r358923)
@@ -309,6 +309,8 @@ should look for the kernel and its modules.
The default is to use the path of the running kernel obtained from the
.Va kern.bootfile
sysctl.
+Modules will also be searched for in /boot/modules if not found in
+.Ar kerneldir .
.It Fl l Ar secs
Set system-wide performance measurement duration for
.Ar secs
___
svn-src-head@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r358924 - head/sys/dev/cxgbe
Author: np
Date: Fri Mar 13 00:12:15 2020
New Revision: 358924
URL: https://svnweb.freebsd.org/changeset/base/358924
Log:
cxgbe(4): Do not display error messages related to the CLIP table if
it's not in use by TOE or KTLS.
Reviewed by: jhb@
MFC after:1 week
Sponsored by: Chelsio Communications
Differential Revision:https://reviews.freebsd.org/D24046
Modified:
head/sys/dev/cxgbe/t4_clip.c
Modified: head/sys/dev/cxgbe/t4_clip.c
==
--- head/sys/dev/cxgbe/t4_clip.cThu Mar 12 23:04:40 2020
(r358923)
+++ head/sys/dev/cxgbe/t4_clip.cFri Mar 13 00:12:15 2020
(r358924)
@@ -273,8 +273,12 @@ update_clip_table(struct adapter *sc)
inet_ntop(AF_INET6, &ce->lip, &ip[0],
sizeof(ip));
- log(LOG_ERR, "%s: could not add %s (%d)\n",
- __func__, ip, rc);
+ if (sc->flags & KERN_TLS_OK ||
+ sc->active_ulds != 0) {
+ log(LOG_ERR,
+ "%s: could not add %s (%d)\n",
+ __func__, ip, rc);
+ }
free(ce, M_CXGBE);
}
next:
___
svn-src-head@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
Re: svn commit: r358923 - in head: lib/libpmcstat usr.sbin/pmcstat
On Thu, Mar 12, 2020 at 6:05 PM Ryan Moeller wrote: > > Author: freqlabs > Date: Thu Mar 12 23:04:40 2020 > New Revision: 358923 > URL: https://svnweb.freebsd.org/changeset/base/358923 > > Log: > libpmcstat: Try /boot/modules if module not found > > Modules from ports/pkg are commonly installed to /boot/modules rather than > to > the same directory the kernel resides in. Look there if a module is not > found > next to the kernel. > > Submitted by: mmacy > Reported by: Nick Principe > Approved by: mmacy (mentor) > MFC after:2 weeks > Sponsored by: iXsystems, Inc. > It's unclear from the context- is there any particular reason this can't instead using sysctl kern.module_path and accept a semicolon delimited list of directories for -k (maybe calling it -k kernelpath instead)? Thanks, Kyle Evans ___ svn-src-head@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r358928 - in head/sys/powerpc: booke powerpc
Author: jhibbits
Date: Fri Mar 13 01:27:37 2020
New Revision: 358928
URL: https://svnweb.freebsd.org/changeset/base/358928
Log:
powerpc: Simplify _nodrop variants of FPU and vector register saves
No need for an extra temporary. It doesn't even help with readability.
Suggested by: kib (almost 2 years ago)
Modified:
head/sys/powerpc/booke/spe.c
head/sys/powerpc/powerpc/altivec.c
head/sys/powerpc/powerpc/fpu.c
Modified: head/sys/powerpc/booke/spe.c
==
--- head/sys/powerpc/booke/spe.cFri Mar 13 01:06:08 2020
(r358927)
+++ head/sys/powerpc/booke/spe.cFri Mar 13 01:27:37 2020
(r358928)
@@ -183,14 +183,11 @@ save_vec(struct thread *td)
void
save_vec_nodrop(struct thread *td)
{
- struct thread *vtd;
struct pcb *pcb;
int i;
- vtd = PCPU_GET(vecthread);
- if (td == vtd) {
+ if (td == PCPU_GET(vecthread))
save_vec_int(td);
- }
pcb = td->td_pcb;
Modified: head/sys/powerpc/powerpc/altivec.c
==
--- head/sys/powerpc/powerpc/altivec.c Fri Mar 13 01:06:08 2020
(r358927)
+++ head/sys/powerpc/powerpc/altivec.c Fri Mar 13 01:27:37 2020
(r358928)
@@ -168,12 +168,7 @@ save_vec(struct thread *td)
void
save_vec_nodrop(struct thread *td)
{
- struct thread *vtd;
- vtd = PCPU_GET(vecthread);
- if (td != vtd) {
- return;
- }
-
- save_vec_int(td);
+ if (td == PCPU_GET(vecthread))
+ save_vec_int(td);
}
Modified: head/sys/powerpc/powerpc/fpu.c
==
--- head/sys/powerpc/powerpc/fpu.c Fri Mar 13 01:06:08 2020
(r358927)
+++ head/sys/powerpc/powerpc/fpu.c Fri Mar 13 01:27:37 2020
(r358928)
@@ -204,12 +204,7 @@ save_fpu(struct thread *td)
void
save_fpu_nodrop(struct thread *td)
{
- struct thread *ftd;
- ftd = PCPU_GET(fputhread);
- if (td != ftd) {
- return;
- }
-
- save_fpu_int(td);
+ if (td == PCPU_GET(fputhread))
+ save_fpu_int(td);
}
___
svn-src-head@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
Re: svn commit: r358923 - in head: lib/libpmcstat usr.sbin/pmcstat
On Thu, Mar 12, 2020, 6:35 PM Kyle Evans wrote: > On Thu, Mar 12, 2020 at 6:05 PM Ryan Moeller wrote: > > > > Author: freqlabs > > Date: Thu Mar 12 23:04:40 2020 > > New Revision: 358923 > > URL: https://svnweb.freebsd.org/changeset/base/358923 > > > > Log: > > libpmcstat: Try /boot/modules if module not found > > > > Modules from ports/pkg are commonly installed to /boot/modules rather > than to > > the same directory the kernel resides in. Look there if a module is > not found > > next to the kernel. > > > > Submitted by: mmacy > > Reported by: Nick Principe > > Approved by: mmacy (mentor) > > MFC after:2 weeks > > Sponsored by: iXsystems, Inc. > > > > It's unclear from the context- is there any particular reason this > can't instead using sysctl kern.module_path and accept a semicolon > delimited list of directories for -k (maybe calling it -k kernelpath > instead)? > There are also #defines for these paths. Warner Thanks, > > Kyle Evans > ___ svn-src-head@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
Re: svn commit: r358923 - in head: lib/libpmcstat usr.sbin/pmcstat
> It's unclear from the context- is there any particular reason this > can't instead using sysctl kern.module_path and accept a semicolon > delimited list of directories for -k (maybe calling it -k kernelpath > instead)? No, that looks technically possible. -Ryan ___ svn-src-head@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/svn-src-head To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
