Re: svn commit: r211505 - head/contrib/gcc
On 20 Aug 2010, at 21:36, Bruce Evans wrote:
> On Fri, 20 Aug 2010, Dimitry Andric wrote:
>
>> On 2010-08-20 00:20, Bruce Evans wrote:
>>> These seem to be needed, and some of them valid. Any lvalue arg that
>>> can be put in a register can be cast to USItype (unsigned int) on i386.
>>> The args are macro args, so they may have any integer type no larger
>>> than USItype originally, and they must be cast to USItype for the asms
>>> to work if the args are smaller than USItype...
>>
>> But will the casts not potentially hide problems, if you pass the wrong
>> types to those macros? Maybe it is better if the compiler complains
>> that some argument is of an incompatible type, than just forcing it to
>> cast?
>
> This is unclear. All integer types are compatible to some extent.
> Upcasting them always works and downcasting them works iff the value
> is not changed.
I think he meant that downcasting might not work.
>
which are apparently "heinous" GNU extensions, so clang can
compile this without using the -fheinous-gnu-extensions option.
>>>
>>> But when the args are lvalues, casting them is invalid. This is
>>> apparently the heinous extension depended on.
>>
>> Yes, clang complains precisely about that:
>>
>> gnu/lib/libgcc/../../../contrib/gcc/libgcc2.c:536:22: error: invalid use of
>> a cast in a inline asm context requiring an l-value: remove the cast or
>> build with -fheinous-gnu-extensions
>> DWunion w = {.ll = __umulsidi3 (uu.s.low, vv.s.low)};
>>^~~~
>> In file included from gnu/lib/libgcc/../../../contrib/gcc/libgcc2.c:65:
>> In file included from gnu/lib/libgcc/../../../contrib/gcc/libgcc2.h:435:
>> gnu/lib/libgcc/../../../contrib/gcc/longlong.h:1293:5: note: instantiated
>> from:
>> umul_ppmm (__w.s.high, __w.s.low, u, v);\
>> ^
>>
>> It turns out that only removing the casts for these specific lvalues is
>> indeed enough to make clang happy. Attached patch reverts all other
>> changes, if that is to be preferred.
>
> I prefer this.
I'll commit this.
>
Results in *no* binary change, neither with clang, nor with gcc.
>>>
>>> This cannot be tested by compiling a few binaries, since the few binaries
>>> might not include enough examples to give test coverage of cases with
>>> args smaller than USItype. Perhaps the macros are only used for building
>>> libgcc, but this is unclear.
>>
>> contrib/gcc/longlong.h is only used in contrib/gcc/libgcc2.h, and in
>> contrib/gcc/config/soft-fp/soft-fp.h. On i386, soft-fp is not used,
>> and libgcc2.h is only used in contrib/gcc/libgcc2.c, so libgcc is the
>> only consumer, as far as I can see.
>
> There are many parameters. Probably the casts are needed for some arches,
> so gnu can't just remove them and wouldn't like your patch. But they should
> fix the casts of lvalues someday.
>
> The ifdefs are hard to untangle. I thought I found a simple case where
> the cast helps, but actually the ifdefs make the cast have no effect
> although the code is logically wrong. From your patch:
>
> % #define count_leading_zeros(count, x) \
> %do {
> \
> % USItype __cbtmp;
> \
> % __asm__ ("bsrl %1,%0" \
> % -: "=r" (__cbtmp) : "rm" (x)); \
> % +: "=r" (__cbtmp) : "rm" ((USItype) (x))); \
> % (count) = __cbtmp ^ 31;
> \
> %} while (0)
> % #define count_trailing_zeros(count, x) \
>
> This interface is supposed to operate on `UWtype x'. UWtype is UDItype
> on some arches, so casting it to USItype breaks it. However, the breakage
> is only logical since all this is under a __i386__ && W_TYPE_SIZE == 32
> ifdef, which makes UWtype the same width as USItype so the cast has no
> effect if x has the correct type.
>
> The above wouldn't work on amd64 since the correct code is bsrq with no
> cast (the value is 64 bits; the cast would give a 32-bit value; apart from
> breaking the value, this would give invalid asm like "bsrq %eax" if the
> casted value ends up in a register. Handling all the cases is apparently
> too hard, so longlong.h doesn't have any special support for clz on amd64
> and a generic version -- the above is apparently used for __clzsi2() and
> _clzdi2() on i386, but on amd64 a slow loop is used. gcc asm still seems
> to be missing support for writing this correctly ("bsr%[mumble1] %1,%0",
> where %[mumble1] is either q or l (or w or b?) depending on the size of %1).
I agree with you. Thanks for the careful analysis.
Regards,
--
Rui Paulo
___
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...
svn commit: r211537 - head/contrib/gcc
Author: rpaulo
Date: Sat Aug 21 10:31:26 2010
New Revision: 211537
URL: http://svn.freebsd.org/changeset/base/211537
Log:
Revert part of r211505. Some of the removed casts are actually safe so
put them back in again. Also, clang only complained about the lvalue
cast.
Submitted by: Dimitry Andric
Pointed out: bde
Modified:
head/contrib/gcc/longlong.h
Modified: head/contrib/gcc/longlong.h
==
--- head/contrib/gcc/longlong.h Sat Aug 21 08:49:53 2010(r211536)
+++ head/contrib/gcc/longlong.h Sat Aug 21 10:31:26 2010(r211537)
@@ -314,38 +314,38 @@ UDItype __umulsidi3 (USItype, USItype);
#if (defined (__i386__) || defined (__i486__)) && W_TYPE_SIZE == 32
#define add_ss(sh, sl, ah, al, bh, bl) \
__asm__ ("addl %5,%1\n\tadcl %3,%0" \
- : "=r" (sh), \
-"=&r" (sl) \
- : "%0" (ah), \
-"g" (bh), \
-"%1" (al), \
-"g" (bl))
+ : "=r" ((USItype) (sh)), \
+"=&r" ((USItype) (sl)) \
+ : "%0" ((USItype) (ah)), \
+"g" ((USItype) (bh)), \
+"%1" ((USItype) (al)), \
+"g" ((USItype) (bl)))
#define sub_ddmmss(sh, sl, ah, al, bh, bl) \
__asm__ ("subl %5,%1\n\tsbbl %3,%0" \
: "=r" (sh), \
"=&r" (sl) \
- : "0" (ah), \
-"g" (bh), \
-"1" (al), \
-"g" (bl))
+ : "0" ((USItype) (ah)), \
+"g" ((USItype) (bh)), \
+"1" ((USItype) (al)), \
+"g" ((USItype) (bl)))
#define umul_ppmm(w1, w0, u, v) \
__asm__ ("mull %3" \
: "=a" (w0), \
"=d" (w1) \
- : "%0" (u), \
-"rm" (v))
+ : "%0" ((USItype) (u)), \
+"rm" ((USItype) (v)))
#define udiv_qrnnd(q, r, n1, n0, dv) \
__asm__ ("divl %4" \
: "=a" (q), \
"=d" (r) \
- : "0" (n0), \
-"1" (n1), \
-"rm" (dv))
+ : "0" ((USItype) (n0)), \
+"1" ((USItype) (n1)), \
+"rm" ((USItype) (dv)))
#define count_leading_zeros(count, x) \
do { \
USItype __cbtmp; \
__asm__ ("bsrl %1,%0" \
-: "=r" (__cbtmp) : "rm" (x)); \
+: "=r" (__cbtmp) : "rm" ((USItype) (x))); \
(count) = __cbtmp ^ 31;\
} while (0)
#define count_trailing_zeros(count, x) \
___
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: r211540 - head/lib/libelf
Author: rpaulo
Date: Sat Aug 21 10:40:12 2010
New Revision: 211540
URL: http://svn.freebsd.org/changeset/base/211540
Log:
Install a PIC libelf. This will be needed by the DTrace runtime
instrumentation object (dtri.o).
Sponsored by: The FreeBSD Foundation
Modified:
head/lib/libelf/Makefile
Modified: head/lib/libelf/Makefile
==
--- head/lib/libelf/MakefileSat Aug 21 10:39:51 2010(r211539)
+++ head/lib/libelf/MakefileSat Aug 21 10:40:12 2010(r211540)
@@ -55,6 +55,7 @@ SRCS= elf_begin.c
\
${GENSRCS}
INCS= libelf.h gelf.h
+INSTALL_PIC_ARCHIVE=yes
GENSRCS= libelf_fsize.c libelf_msize.c libelf_convert.c
CLEANFILES=${GENSRCS}
CFLAGS+= -I. -I${.CURDIR}
___
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: r211545 - in head/cddl/contrib/opensolaris/cmd/dtrace/test: cmd/scripts tst/common/aggs tst/common/assocs tst/common/dtraceUtil tst/common/funcs tst/common/ip tst/common/misc tst/common...
Author: rpaulo Date: Sat Aug 21 11:04:38 2010 New Revision: 211545 URL: http://svn.freebsd.org/changeset/base/211545 Log: Port most of the DTrace tests to FreeBSD. Sponsored by: The FreeBSD Foundation Added: head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/usdt/Makefile (contents, props changed) head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/usdt/main.c (contents, props changed) head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/usdt/prov.d head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/usdt/prov.h (contents, props changed) Modified: head/cddl/contrib/opensolaris/cmd/dtrace/test/cmd/scripts/dtest.pl head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/aggs/tst.subr.d head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/assocs/tst.orthogonality.d head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/dtraceUtil/tst.AddSearchPath.d.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/dtraceUtil/tst.DestructWithModule.d.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/dtraceUtil/tst.ELFGenerationOut.d.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/dtraceUtil/tst.ELFGenerationWithO.d.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/dtraceUtil/tst.PreprocessorStatement.d.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/funcs/err.D_PROTO_LEN.motoofew.d head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/funcs/err.D_PROTO_LEN.motoomany.d head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/funcs/err.D_PROTO_LEN.mtatoofew.d head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/funcs/err.D_PROTO_LEN.mtatoomany.d head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/funcs/tst.index.d head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/ip/get.ipv4remote.pl head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/ip/get.ipv6remote.pl head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/ip/tst.ipv4localicmp.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/ip/tst.ipv4localtcp.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/ip/tst.ipv4localudp.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/ip/tst.ipv4remoteicmp.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/ip/tst.ipv4remotetcp.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/ip/tst.ipv4remoteudp.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/ip/tst.ipv6localicmp.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/ip/tst.ipv6remoteicmp.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/ip/tst.localtcpstate.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/ip/tst.remotetcpstate.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/misc/tst.include.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/misc/tst.schrock.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/pid/tst.fork.c head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/pid/tst.gcc.c head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/pid/tst.provregex1.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/pid/tst.provregex2.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/pid/tst.provregex3.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/pid/tst.provregex4.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/printa/tst.largeusersym.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/printa/tst.walltimestamp.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/printf/tst.printT.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/printf/tst.printY.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/printf/tst.sym.d head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/printf/tst.sym.d.out head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/proc/tst.exec.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/proc/tst.execfail.ENOENT.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/proc/tst.execfail.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/proc/tst.sigwait.d head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/scripting/tst.D_MACRO_UNUSED.overflow.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/scripting/tst.arguments.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/scripting/tst.egid.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/scripting/tst.euid.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/scripting/tst.gid.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/scripting/tst.ppid.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/scripting/tst.projid.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/scripting/tst.sid.ksh head/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/scr
svn commit: r211546 - head/sys/net80211
Author: bschmidt Date: Sat Aug 21 11:06:21 2010 New Revision: 211546 URL: http://svn.freebsd.org/changeset/base/211546 Log: Wrap remaining ieee80211_ratectl_node_init() calls missed in r211314. MFC after:1 week Modified: head/sys/net80211/ieee80211_node.c Modified: head/sys/net80211/ieee80211_node.c == --- head/sys/net80211/ieee80211_node.c Sat Aug 21 11:04:38 2010 (r211545) +++ head/sys/net80211/ieee80211_node.c Sat Aug 21 11:06:21 2010 (r211546) @@ -1404,7 +1404,8 @@ ieee80211_fakeup_adhoc_node(struct ieee8 #endif } ieee80211_node_setuptxparms(ni); - ieee80211_ratectl_node_init(ni); + if (vap->iv_caps & IEEE80211_C_RATECTL) + ieee80211_ratectl_node_init(ni); if (ic->ic_newassoc != NULL) ic->ic_newassoc(ni, 1); /* XXX not right for 802.1x/WPA */ @@ -1474,7 +1475,8 @@ ieee80211_add_neighbor(struct ieee80211v if (ieee80211_iserp_rateset(&ni->ni_rates)) ni->ni_flags |= IEEE80211_NODE_ERP; ieee80211_node_setuptxparms(ni); - ieee80211_ratectl_node_init(ni); + if (vap->iv_caps & IEEE80211_C_RATECTL) + ieee80211_ratectl_node_init(ni); if (ic->ic_newassoc != NULL) ic->ic_newassoc(ni, 1); /* XXX not right for 802.1x/WPA */ @@ -2343,7 +2345,8 @@ ieee80211_node_join(struct ieee80211_nod ); ieee80211_node_setuptxparms(ni); - ieee80211_ratectl_node_init(ni); + if (vap->iv_caps & IEEE80211_C_RATECTL) + ieee80211_ratectl_node_init(ni); /* give driver a chance to setup state like ni_txrate */ if (ic->ic_newassoc != NULL) ic->ic_newassoc(ni, newassoc); ___ 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: r211547 - head/cddl/lib
Author: rpaulo
Date: Sat Aug 21 11:09:03 2010
New Revision: 211547
URL: http://svn.freebsd.org/changeset/base/211547
Log:
The DTrace instrumentation object is going to be i386/amd64 only.
Sponsored by: The FreeBSD Foundation
Modified:
head/cddl/lib/Makefile
Modified: head/cddl/lib/Makefile
==
--- head/cddl/lib/Makefile Sat Aug 21 11:06:21 2010(r211546)
+++ head/cddl/lib/Makefile Sat Aug 21 11:09:03 2010(r211547)
@@ -2,7 +2,7 @@
.include
-SUBDIR=drti \
+SUBDIR=${_drti} \
libavl \
libctf \
${_libdtrace} \
@@ -20,6 +20,7 @@ _libzpool=libzpool
.endif
.if ${MACHINE_ARCH} == amd64 || ${MACHINE_ARCH} == i386
+_drti= drti
_libdtrace=libdtrace
.endif
___
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: r211550 - head/cddl/lib
Author: rpaulo
Date: Sat Aug 21 11:14:50 2010
New Revision: 211550
URL: http://svn.freebsd.org/changeset/base/211550
Log:
Use double quotes when checking the value of MACHINE_ARCH.
Sponsored by: The FreeBSD Foundation
Modified:
head/cddl/lib/Makefile
Modified: head/cddl/lib/Makefile
==
--- head/cddl/lib/Makefile Sat Aug 21 11:11:32 2010(r211549)
+++ head/cddl/lib/Makefile Sat Aug 21 11:14:50 2010(r211550)
@@ -19,7 +19,7 @@ _libzpool=libzpool
.endif
.endif
-.if ${MACHINE_ARCH} == amd64 || ${MACHINE_ARCH} == i386
+.if ${MACHINE_ARCH} == "amd64" || ${MACHINE_ARCH} == "i386"
_drti= drti
_libdtrace=libdtrace
.endif
___
svn-src-head@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail to "svn-src-head-unsubscr...@freebsd.org"
svn commit: r211553 - head/sys/cddl/compat/opensolaris/kern
Author: rpaulo
Date: Sat Aug 21 11:41:32 2010
New Revision: 211553
URL: http://svn.freebsd.org/changeset/base/211553
Log:
Add sysname to struct opensolaris_utsname. This is needed by one DTrace
test.
Sponsored by: The FreeBSD Foundation
Modified:
head/sys/cddl/compat/opensolaris/kern/opensolaris_misc.c
Modified: head/sys/cddl/compat/opensolaris/kern/opensolaris_misc.c
==
--- head/sys/cddl/compat/opensolaris/kern/opensolaris_misc.cSat Aug 21
11:33:49 2010(r211552)
+++ head/sys/cddl/compat/opensolaris/kern/opensolaris_misc.cSat Aug 21
11:41:32 2010(r211553)
@@ -38,7 +38,8 @@ __FBSDID("$FreeBSD$");
char hw_serial[11] = "0";
struct opensolaris_utsname utsname = {
- .nodename = "unset"
+ .nodename = "unset",
+ .sysname = "SunOS"
};
int
___
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: r211554 - in head/cddl: contrib/opensolaris/cmd/dtrace contrib/opensolaris/lib/libdtrace/common contrib/opensolaris/lib/libdtrace/i386 lib/libdtrace
Author: rpaulo
Date: Sat Aug 21 11:50:53 2010
New Revision: 211554
URL: http://svn.freebsd.org/changeset/base/211554
Log:
Add libdtrace support for tracing userland programs.
Summary of changes:
* Implement a compatibility shim between Solaris libproc and our
libproc and remove several ifdefs because of this.
* Port the drti to FreeBSD.
* Implement the missing DOODAD sections
* Link with libproc and librtld_db
* Support for ustack, jstack and uregs (by sson@)
* Misc bugfixing
When writing the SUWN_dof section, we had to resort to building the ELF
file layout by "hand". This is the job of libelf, but our libelf doesn't
support this yet. When libelf is fixed, we can remove the code under
#ifdef BROKEN_LIBELF.
Sponsored by: The FreeBSD Foundation
Added:
head/cddl/lib/libdtrace/libproc_compat.h (contents, props changed)
head/cddl/lib/libdtrace/regs_x86.d (contents, props changed)
Modified:
head/cddl/contrib/opensolaris/cmd/dtrace/dtrace.c
head/cddl/contrib/opensolaris/lib/libdtrace/common/drti.c
head/cddl/contrib/opensolaris/lib/libdtrace/common/dt_aggregate.c
head/cddl/contrib/opensolaris/lib/libdtrace/common/dt_consume.c
head/cddl/contrib/opensolaris/lib/libdtrace/common/dt_link.c
head/cddl/contrib/opensolaris/lib/libdtrace/common/dt_open.c
head/cddl/contrib/opensolaris/lib/libdtrace/common/dt_pid.c
head/cddl/contrib/opensolaris/lib/libdtrace/common/dt_proc.c
head/cddl/contrib/opensolaris/lib/libdtrace/common/dt_proc.h
head/cddl/contrib/opensolaris/lib/libdtrace/common/dt_subr.c
head/cddl/contrib/opensolaris/lib/libdtrace/common/dtrace.h
head/cddl/contrib/opensolaris/lib/libdtrace/i386/dt_isadep.c
head/cddl/lib/libdtrace/Makefile
Modified: head/cddl/contrib/opensolaris/cmd/dtrace/dtrace.c
==
--- head/cddl/contrib/opensolaris/cmd/dtrace/dtrace.c Sat Aug 21 11:41:32
2010(r211553)
+++ head/cddl/contrib/opensolaris/cmd/dtrace/dtrace.c Sat Aug 21 11:50:53
2010(r211554)
@@ -773,19 +773,27 @@ compile_str(dtrace_cmd_t *dcp)
static void
prochandler(struct ps_prochandle *P, const char *msg, void *arg)
{
-fatal("DOODAD in function %s, file %s, line
%d\n",__FUNCTION__,__FILE__,__LINE__);
-#ifdef DOODAD
+#if defined(sun)
const psinfo_t *prp = Ppsinfo(P);
int pid = Pstatus(P)->pr_pid;
char name[SIG2STR_MAX];
+#else
+ int wstatus = proc_getwstat(P);
+ int pid = proc_getpid(P);
+#endif
if (msg != NULL) {
notice("pid %d: %s\n", pid, msg);
return;
}
+#if defined(sun)
switch (Pstate(P)) {
+#else
+ switch (proc_state(P)) {
+#endif
case PS_UNDEAD:
+#if defined(sun)
/*
* Ideally we would like to always report pr_wstat here, but it
* isn't possible given current /proc semantics. If we grabbed
@@ -798,9 +806,20 @@ fatal("DOODAD in function %s, file %s, l
notice("pid %d terminated by %s\n", pid,
proc_signame(WTERMSIG(prp->pr_wstat),
name, sizeof (name)));
+#else
+ if (WIFSIGNALED(wstatus)) {
+ notice("pid %d terminated by %d\n", pid,
+ WTERMSIG(wstatus));
+#endif
+#if defined(sun)
} else if (prp != NULL && WEXITSTATUS(prp->pr_wstat) != 0) {
notice("pid %d exited with status %d\n",
pid, WEXITSTATUS(prp->pr_wstat));
+#else
+ } else if (WEXITSTATUS(wstatus) != 0) {
+ notice("pid %d exited with status %d\n",
+ pid, WEXITSTATUS(wstatus));
+#endif
} else {
notice("pid %d has exited\n", pid);
}
@@ -812,7 +831,6 @@ fatal("DOODAD in function %s, file %s, l
g_pslive--;
break;
}
-#endif
}
/*ARGSUSED*/
Modified: head/cddl/contrib/opensolaris/lib/libdtrace/common/drti.c
==
--- head/cddl/contrib/opensolaris/lib/libdtrace/common/drti.c Sat Aug 21
11:41:32 2010(r211553)
+++ head/cddl/contrib/opensolaris/lib/libdtrace/common/drti.c Sat Aug 21
11:50:53 2010(r211554)
@@ -34,6 +34,8 @@
#include
#include
#include
+#include
+#include
/*
* In Solaris 10 GA, the only mechanism for communicating helper information
@@ -53,12 +55,16 @@
*/
static const char *devnamep = "/dev/dtrace/helper";
+#if defined(sun)
static const char *olddevname = "/devices/pseudo/dtr...@0:helper";
+#endif
static const char *modname;/* Name of this load object */
static int gen;/* DOF helper generation */
+#if defined(sun)
extern dof_hdr_t __SUNW_dof; /* DOF defined in the .SUNW_dof section */
-static boolean_t dof_init_debug = B_FALSE;
svn commit: r211555 - head/sys/cddl/contrib/opensolaris/uts/common/sys
Author: rpaulo
Date: Sat Aug 21 11:58:08 2010
New Revision: 211555
URL: http://svn.freebsd.org/changeset/base/211555
Log:
Port the DTrace helper ioctls to FreeBSD and add a helper member to
dof_helper_t (needed by drti.o).
Sponsored by: The FreeBSD Foundation
Modified:
head/sys/cddl/contrib/opensolaris/uts/common/sys/dtrace.h
Modified: head/sys/cddl/contrib/opensolaris/uts/common/sys/dtrace.h
==
--- head/sys/cddl/contrib/opensolaris/uts/common/sys/dtrace.h Sat Aug 21
11:50:53 2010(r211554)
+++ head/sys/cddl/contrib/opensolaris/uts/common/sys/dtrace.h Sat Aug 21
11:58:08 2010(r211555)
@@ -1322,15 +1322,24 @@ typedef struct {
* helpers and should no longer be used. No other ioctls are valid on the
* helper minor node.
*/
+#if defined(sun)
#defineDTRACEHIOC (('d' << 24) | ('t' << 16) | ('h' << 8))
#defineDTRACEHIOC_ADD (DTRACEHIOC | 1)/* add helper */
#defineDTRACEHIOC_REMOVE (DTRACEHIOC | 2)/* remove
helper */
#defineDTRACEHIOC_ADDDOF (DTRACEHIOC | 3)/* add helper
DOF */
+#else
+#defineDTRACEHIOC_ADD _IOWR('z', 1, dof_hdr_t)/* add helper */
+#defineDTRACEHIOC_REMOVE _IOW('z', 2, int) /* remove
helper */
+#defineDTRACEHIOC_ADDDOF _IOWR('z', 3, dof_helper_t)/* add
helper DOF */
+#endif
typedef struct dof_helper {
char dofhp_mod[DTRACE_MODNAMELEN]; /* executable or library name */
uint64_t dofhp_addr;/* base address of object */
uint64_t dofhp_dof; /* address of helper DOF */
+#if !defined(sun)
+ int gen;
+#endif
} dof_helper_t;
#defineDTRACEMNR_DTRACE"dtrace"/* node for DTrace ops
*/
@@ -2219,10 +2228,11 @@ extern void dtrace_vtime_enable(void);
extern void dtrace_vtime_disable(void);
struct regs;
+struct reg;
#if defined(sun)
-extern int (*dtrace_pid_probe_ptr)(struct regs *);
-extern int (*dtrace_return_probe_ptr)(struct regs *);
+extern int (*dtrace_pid_probe_ptr)(struct reg *);
+extern int (*dtrace_return_probe_ptr)(struct reg *);
extern void (*dtrace_fasttrap_fork_ptr)(proc_t *, proc_t *);
extern void (*dtrace_fasttrap_exec_ptr)(proc_t *);
extern void (*dtrace_fasttrap_exit_ptr)(proc_t *);
___
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: r211501 - in head/sys: modules modules/send netinet netinet6 sys
On 20 Aug 2010, at 02:33, Doug Barton wrote: > On 08/19/2010 14:50, Bjoern A. Zeeb wrote: >> On Thu, 19 Aug 2010, Doug Barton wrote: >> >>> On 08/19/2010 04:42, Ana Kukec wrote: Hi all, A man page for the kernel side SeND, and an updated port with Secure Neighbor Discovery application will follow soon. >>> >>> At minimum committing the man page along with new code is how such >>> things should be done. Can you define "soon?" >> >> Yes I can: soon will be when it'll be comitted. > > I'm guessing that's an attempt at humor, but what I'm looking for is an > actual timeline. If the answer is "in the next day or two" then that's fine, > we'll let this one go with a warning. If the answer is "someday" then I'm > going to request that the code be backed out until the man page(s) are done. > I'm actually trying to be reasonable and work with a new developer so that > they can learn how we like to have things done around here. What are you > trying to do? What? Are you serious? Do you really want some code to be backed out because it doesn't have a man page yet? Talking about overkill.. You don't seem to have an idea how many times code was committed without a man page. I didn't see you complain back then.. It's very likely you wouldn't be complaining if the developer in question hadn't posted a message to the mailing list saying she intends to commit a man page soon. > >> Meanwhile if you have time I'd know several dozen other things we >> don't have a man page for ... > > I would definitely recommend that you make a list of them, preferably > someplace visible like the wiki. If you don't already have access to the wiki > (or some other suitable forum) then send the list to -doc@ and we'll make > sure that it gets posted. Given the recent resurgence of interest in updating > our documentation it certainly wouldn't hurt to have some examples of > low-hanging fruit that we can turn budding mdoc gurus loose on. > > Meanwhile, the fact that in the past code has been committed without man > pages is the exact cause of the problem you're describing. It is certainly > not a justification for continuing to do it wrong. Reverting code just because it doesn't have a man page is pointless. Everyone can write the damn man page, but not everyone can write code. You should really take this into consideration when proposing a commit to be reverted. Regards, -- Rui Paulo ___ 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: r211558 - head/cddl/contrib/dtracetoolkit
Author: rpaulo Date: Sat Aug 21 13:21:05 2010 New Revision: 211558 URL: http://svn.freebsd.org/changeset/base/211558 Log: Add a directory for the DTraceToolkit. Sponsored by: The FreeBSD Foundation Added: head/cddl/contrib/dtracetoolkit/ ___ 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: r211560 - head
Author: rpaulo
Date: Sat Aug 21 13:42:12 2010
New Revision: 211560
URL: http://svn.freebsd.org/changeset/base/211560
Log:
Add libelf to the prebuild libs.
This is needed for the DTrace instrumentation object.
Sponsored by: The FreeBSD Foundation
Modified:
head/Makefile.inc1
Modified: head/Makefile.inc1
==
--- head/Makefile.inc1 Sat Aug 21 13:23:54 2010(r211559)
+++ head/Makefile.inc1 Sat Aug 21 13:42:12 2010(r211560)
@@ -1142,7 +1142,7 @@ gnu/lib/libgcc__L: lib/libc__L
_prebuild_libs=${_kerberos5_lib_libasn1} ${_kerberos5_lib_libheimntlm}
\
${_kerberos5_lib_libhx509} ${_kerberos5_lib_libkrb5} \
${_kerberos5_lib_libroken} \
- lib/libbz2 lib/libcom_err lib/libcrypt \
+ lib/libbz2 lib/libcom_err lib/libcrypt lib/libelf \
lib/libexpat lib/libfetch \
${_lib_libgssapi} ${_lib_libipx} \
lib/libkiconv lib/libkvm lib/liblzma lib/libmd \
___
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: r211561 - head/cddl/lib/drti
Author: rpaulo
Date: Sat Aug 21 13:48:04 2010
New Revision: 211561
URL: http://svn.freebsd.org/changeset/base/211561
Log:
Link drti.o with a PIC version of libelf. This is needed because
drti.o depends on libelf and this avoids linking every other drti.o
program (namely programs with USDT probes) with libelf.
Sponsored by: The FreeBSD Foundation
Modified:
head/cddl/lib/drti/Makefile
Modified: head/cddl/lib/drti/Makefile
==
--- head/cddl/lib/drti/Makefile Sat Aug 21 13:42:12 2010(r211560)
+++ head/cddl/lib/drti/Makefile Sat Aug 21 13:48:04 2010(r211561)
@@ -8,13 +8,20 @@ FILESOWN= ${LIBOWN}
FILESGRP= ${LIBGRP}
FILESMODE= ${LIBMODE}
FILESDIR= ${LIBDIR}/dtrace
-CLEANFILES=${FILES}
+CLEANFILES=${FILES} ${FILES}.tmp
CFLAGS+= -I${.CURDIR}/../../../sys/cddl/compat/opensolaris \
-I${.CURDIR}/../../../cddl/compat/opensolaris/include \
-I${OPENSOLARIS_USR_DISTDIR}/head \
-I${OPENSOLARIS_USR_DISTDIR}/lib/libctf/common \
-I${OPENSOLARIS_USR_DISTDIR}/lib/libdtrace/common \
- -I${OPENSOLARIS_SYS_DISTDIR}/uts/common
+ -I${OPENSOLARIS_SYS_DISTDIR}/uts/common \
+ -DPIC -fpic
+
+.c.o:
+ ${CC} ${CFLAGS} -c ${.IMPSRC}
+ mv ${FILES} ${FILES}.tmp
+ ${LD} -o ${FILES} -r ${FILES}.tmp ${LIBELF_PIC}
+ rm -f ${FILES}.tmp
.include
___
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: r211562 - in head/cddl: contrib/opensolaris/cmd/plockstat usr.sbin/plockstat
Author: rpaulo
Date: Sat Aug 21 13:54:33 2010
New Revision: 211562
URL: http://svn.freebsd.org/changeset/base/211562
Log:
Port plockstat to FreeBSD. This will be connected to the build later.
Sponsored by: The FreeBSD Foundation
Added:
head/cddl/usr.sbin/plockstat/
head/cddl/usr.sbin/plockstat/Makefile (contents, props changed)
Modified:
head/cddl/contrib/opensolaris/cmd/plockstat/plockstat.c
Modified: head/cddl/contrib/opensolaris/cmd/plockstat/plockstat.c
==
--- head/cddl/contrib/opensolaris/cmd/plockstat/plockstat.c Sat Aug 21
13:48:04 2010(r211561)
+++ head/cddl/contrib/opensolaris/cmd/plockstat/plockstat.c Sat Aug 21
13:54:33 2010(r211562)
@@ -24,7 +24,9 @@
* Use is subject to license terms.
*/
+#if defined(sun)
#pragma ident "%Z%%M% %I% %E% SMI"
+#endif
#include
#include
@@ -41,6 +43,7 @@
#include
#include
#include
+#include
static char *g_pname;
static dtrace_hdl_t *g_dtp;
@@ -502,7 +505,12 @@ getsym(struct ps_prochandle *P, uintptr_
{
char name[256];
GElf_Sym sym;
+#if defined(sun)
prsyminfo_t info;
+#else
+ prmap_t *map;
+ int info; /* XXX unused */
+#endif
size_t len;
if (P == NULL || Pxlookup_by_addr(P, addr, name, sizeof (name),
@@ -510,6 +518,7 @@ getsym(struct ps_prochandle *P, uintptr_
(void) snprintf(buf, size, "%#lx", addr);
return (0);
}
+#if defined(sun)
if (info.prs_object == NULL)
info.prs_object = "";
@@ -520,15 +529,19 @@ getsym(struct ps_prochandle *P, uintptr_
}
len = snprintf(buf, size, "%s`%s", info.prs_object, info.prs_name);
+#else
+ map = proc_addr2map(P, addr);
+ len = snprintf(buf, size, "%s`%s", map->pr_mapname, name);
+#endif
buf += len;
size -= len;
if (sym.st_value != addr)
len = snprintf(buf, size, "+%#lx", addr - sym.st_value);
- if (nolocks && strcmp("libc.so.1", info.prs_object) == 0 &&
- (strstr("mutex", info.prs_name) == 0 ||
- strstr("rw", info.prs_name) == 0))
+ if (nolocks && strcmp("libc.so.1", map->pr_mapname) == 0 &&
+ (strstr("mutex", name) == 0 ||
+ strstr("rw", name) == 0))
return (-1);
return (0);
@@ -655,8 +668,13 @@ process_aggregate(const dtrace_aggdata_t
static void
prochandler(struct ps_prochandle *P, const char *msg, void *arg)
{
+#if defined(sun)
const psinfo_t *prp = Ppsinfo(P);
int pid = Pstatus(P)->pr_pid;
+#else
+ int pid = proc_getpid(P);
+ int wstat = proc_getwstat(P);
+#endif
char name[SIG2STR_MAX];
if (msg != NULL) {
@@ -674,13 +692,13 @@ prochandler(struct ps_prochandle *P, con
* When /proc provides a stable pr_wstat in the status file,
* this code can be improved by examining this new pr_wstat.
*/
- if (prp != NULL && WIFSIGNALED(prp->pr_wstat)) {
+ if (WIFSIGNALED(wstat)) {
notice("pid %d terminated by %s\n", pid,
- proc_signame(WTERMSIG(prp->pr_wstat),
+ proc_signame(WTERMSIG(wstat),
name, sizeof (name)));
- } else if (prp != NULL && WEXITSTATUS(prp->pr_wstat) != 0) {
+ } else if (WEXITSTATUS(wstat) != 0) {
notice("pid %d exited with status %d\n",
- pid, WEXITSTATUS(prp->pr_wstat));
+ pid, WEXITSTATUS(wstat));
} else {
notice("pid %d has exited\n", pid);
}
@@ -755,7 +773,9 @@ intr(int signo)
int
main(int argc, char **argv)
{
+#if defined(sun)
ucred_t *ucp;
+#endif
int err;
int opt_C = 0, opt_H = 0, opt_p = 0, opt_v = 0;
char c, *p, *end;
@@ -764,7 +784,7 @@ main(int argc, char **argv)
g_pname = basename(argv[0]);
argv[0] = g_pname; /* rewrite argv[0] for getopt errors */
-
+#if defined(sun)
/*
* Make sure we have the required dtrace_proc privilege.
*/
@@ -777,6 +797,7 @@ main(int argc, char **argv)
ucred_free(ucp);
}
+#endif
while ((c = getopt(argc, argv, PLOCKSTAT_OPTSTR)) != EOF) {
switch (c) {
@@ -907,7 +928,7 @@ main(int argc, char **argv)
if (opt_H) {
dprog_add(g_hold_init);
- if (g_opt_s == NULL)
+ if (!g_opt_s)
dprog_add(g_hold_times);
else
dprog_add(g_hold_histogram);
@@ -915,7 +936,7 @@ main(int argc, char **argv)
if (opt_C) {
dprog_add(g_ctnd_init);
- if (g_opt_s == NULL)
+ if (!g_opt_s)
dprog_add(g_ctnd_tim
svn commit: r211563 - head/cddl/usr.sbin/dtruss
Author: rpaulo
Date: Sat Aug 21 13:56:29 2010
New Revision: 211563
URL: http://svn.freebsd.org/changeset/base/211563
Log:
Add a Makefile for dtruss.
Sponsored by: The FreeBSD Foundation
Added:
head/cddl/usr.sbin/dtruss/
head/cddl/usr.sbin/dtruss/Makefile (contents, props changed)
Added: head/cddl/usr.sbin/dtruss/Makefile
==
--- /dev/null 00:00:00 1970 (empty, because file is newly added)
+++ head/cddl/usr.sbin/dtruss/Makefile Sat Aug 21 13:56:29 2010
(r211563)
@@ -0,0 +1,8 @@
+# $FreeBSD$
+
+.PATH: ${.CURDIR}/../../../cddl/contrib/dtracetoolkit
+
+SCRIPTS=dtruss
+NO_MAN=
+
+.include
___
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: r211561 - head/cddl/lib/drti
On Sat, Aug 21, 2010 at 01:48:05PM +, Rui Paulo wrote:
> Author: rpaulo
> Date: Sat Aug 21 13:48:04 2010
> New Revision: 211561
> URL: http://svn.freebsd.org/changeset/base/211561
>
> Log:
> Link drti.o with a PIC version of libelf. This is needed because
> drti.o depends on libelf and this avoids linking every other drti.o
> program (namely programs with USDT probes) with libelf.
What is wrong with having libelf as dependency of the resulting
binary ?
>
> Sponsored by: The FreeBSD Foundation
>
> Modified:
> head/cddl/lib/drti/Makefile
>
> Modified: head/cddl/lib/drti/Makefile
> ==
> --- head/cddl/lib/drti/Makefile Sat Aug 21 13:42:12 2010
> (r211560)
> +++ head/cddl/lib/drti/Makefile Sat Aug 21 13:48:04 2010
> (r211561)
> @@ -8,13 +8,20 @@ FILESOWN= ${LIBOWN}
> FILESGRP=${LIBGRP}
> FILESMODE= ${LIBMODE}
> FILESDIR=${LIBDIR}/dtrace
> -CLEANFILES= ${FILES}
> +CLEANFILES= ${FILES} ${FILES}.tmp
>
> CFLAGS+= -I${.CURDIR}/../../../sys/cddl/compat/opensolaris \
> -I${.CURDIR}/../../../cddl/compat/opensolaris/include \
> -I${OPENSOLARIS_USR_DISTDIR}/head \
> -I${OPENSOLARIS_USR_DISTDIR}/lib/libctf/common \
> -I${OPENSOLARIS_USR_DISTDIR}/lib/libdtrace/common \
> - -I${OPENSOLARIS_SYS_DISTDIR}/uts/common
> + -I${OPENSOLARIS_SYS_DISTDIR}/uts/common \
> + -DPIC -fpic
> +
> +.c.o:
> + ${CC} ${CFLAGS} -c ${.IMPSRC}
> + mv ${FILES} ${FILES}.tmp
> + ${LD} -o ${FILES} -r ${FILES}.tmp ${LIBELF_PIC}
> + rm -f ${FILES}.tmp
>
> .include
pgp4mvc55W276.pgp
Description: PGP signature
Re: svn commit: r211505 - head/contrib/gcc
On 2010-08-20 22:36, Bruce Evans wrote:
> On Fri, 20 Aug 2010, Dimitry Andric wrote:
[...]
>> But will the casts not potentially hide problems, if you pass the wrong
>> types to those macros? Maybe it is better if the compiler complains
>> that some argument is of an incompatible type, than just forcing it to
>> cast?
> This is unclear. All integer types are compatible to some extent.
> Upcasting them always works and downcasting them works iff the value
> is not changed.
I meant this in the context of this llvm PR, about matching inline asm
input constraints with output constraints of an incompatible type:
http://llvm.org/bugs/show_bug.cgi?id=3373
Clang is currently somewhat pickier about the arguments to inline asm,
which we also noticed in OpenSSL code, where a rotate-left macro is
defined (for i386 and amd64) as:
# define ROTATE(a,n) ({ register unsigned int ret; \
asm ( \
"roll %1,%0" \
: "=r"(ret) \
: "I"(n), "0"(a) \
: "cc"); \
ret; \
})
On amd64, it was being called with the 'a' argument being of unsigned
long type. Clang complained:
crypto/openssl/crypto/md4/md4_dgst.c:117:2:
error: unsupported inline asm: input with type 'unsigned long' matching
output with type 'unsigned int'
R0(A,B,C,D,X( 0), 3,0); HOST_c2l(data,l); X( 2)=l;
^~
In this case, the OpenSSL developers chose to explicitly cast 'a' to
'unsigned int' (see ).
___
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: r211564 - head/cddl/usr.sbin
Author: rpaulo
Date: Sat Aug 21 14:02:04 2010
New Revision: 211564
URL: http://svn.freebsd.org/changeset/base/211564
Log:
Use double quotes when checking the value of MACHINE_ARCH.
Sponsored by: The FreeBSD Foundation
Modified:
head/cddl/usr.sbin/Makefile
Modified: head/cddl/usr.sbin/Makefile
==
--- head/cddl/usr.sbin/Makefile Sat Aug 21 13:56:29 2010(r211563)
+++ head/cddl/usr.sbin/Makefile Sat Aug 21 14:02:04 2010(r211564)
@@ -12,7 +12,7 @@ _zdb= zdb
.endif
.endif
-.if ${MACHINE_ARCH} == amd64 || ${MACHINE_ARCH} == i386
+.if ${MACHINE_ARCH} == "amd64" || ${MACHINE_ARCH} == "i386"
_dtrace= dtrace
_lockstat= lockstat
.endif
___
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: r211565 - head/share/mk
Author: rpaulo
Date: Sat Aug 21 14:08:04 2010
New Revision: 211565
URL: http://svn.freebsd.org/changeset/base/211565
Log:
Add the definition of LIBELF_PIC.
Sponsored by: The FreeBSD Foundation
Modified:
head/share/mk/bsd.libnames.mk
Modified: head/share/mk/bsd.libnames.mk
==
--- head/share/mk/bsd.libnames.mk Sat Aug 21 14:02:04 2010
(r211564)
+++ head/share/mk/bsd.libnames.mk Sat Aug 21 14:08:04 2010
(r211565)
@@ -45,6 +45,7 @@ LIBDTRACE?= ${DESTDIR}${LIBDIR}/libdtrac
LIBDWARF?= ${DESTDIR}${LIBDIR}/libdwarf.a
LIBEDIT?= ${DESTDIR}${LIBDIR}/libedit.a
LIBELF?= ${DESTDIR}${LIBDIR}/libelf.a
+LIBELF_PIC?= ${DESTDIR}${LIBDIR}/libelf_pic.a
LIBFETCH?= ${DESTDIR}${LIBDIR}/libfetch.a
LIBFL?="don't use LIBFL, use LIBL"
LIBFORM?= ${DESTDIR}${LIBDIR}/libform.a
___
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: r211566 - head/sys/cddl/compat/opensolaris/sys
Author: rpaulo
Date: Sat Aug 21 14:09:24 2010
New Revision: 211566
URL: http://svn.freebsd.org/changeset/base/211566
Log:
Add a sysname char * to struct opensolaris_utsname.
Sponsored by: The FreeBSD Foundation
Modified:
head/sys/cddl/compat/opensolaris/sys/misc.h
Modified: head/sys/cddl/compat/opensolaris/sys/misc.h
==
--- head/sys/cddl/compat/opensolaris/sys/misc.h Sat Aug 21 14:08:04 2010
(r211565)
+++ head/sys/cddl/compat/opensolaris/sys/misc.h Sat Aug 21 14:09:24 2010
(r211566)
@@ -46,6 +46,7 @@
#ifdef _KERNEL
struct opensolaris_utsname {
char *nodename;
+ char *sysname;
};
extern char hw_serial[11];
___
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: r211567 - head/tools/test/dtrace
Author: rpaulo
Date: Sat Aug 21 14:14:24 2010
New Revision: 211567
URL: http://svn.freebsd.org/changeset/base/211567
Log:
Update for the new tests.
Sponsored by: The FreeBSD Foundation
Modified:
head/tools/test/dtrace/Makefile
Modified: head/tools/test/dtrace/Makefile
==
--- head/tools/test/dtrace/Makefile Sat Aug 21 14:09:24 2010
(r211566)
+++ head/tools/test/dtrace/Makefile Sat Aug 21 14:14:24 2010
(r211567)
@@ -1,274 +1,206 @@
# $FreeBSD$
TESTSRCDIR=${.CURDIR}/../../../cddl/contrib/opensolaris/cmd/dtrace/test
+CFLAGS+= -w
DTEST= ${TESTSRCDIR}/cmd/scripts/dtest.pl -n
IGNORE=\
- ${TESTSRCDIR}/tst/common/arrays/err.D_IDENT_UNDEF.badureg.d
\
- ${TESTSRCDIR}/tst/common/printa/err.D_PRINTF_ARG_TYPE.jstack.d
\
- ${TESTSRCDIR}/tst/common/printa/err.D_PRINTF_ARG_TYPE.ustack.d
\
- ${TESTSRCDIR}/tst/common/printa/tst.stack.d
\
- ${TESTSRCDIR}/tst/common/stack/err.D_USTACK_FRAMES.bad.d
\
- ${TESTSRCDIR}/tst/common/stack/err.D_USTACK_PROTO.bad.d
\
- ${TESTSRCDIR}/tst/common/stack/err.D_USTACK_STRSIZE.bad.d
\
- ${TESTSRCDIR}/tst/common/stop/tst.stop1.d
\
- ${TESTSRCDIR}/tst/common/aggs/tst.subr.d
\
- ${TESTSRCDIR}/tst/common/aggs/tst.signature.d
\
- ${TESTSRCDIR}/tst/common/arrays/tst.uregsarray.d
\
- ${TESTSRCDIR}/tst/common/assocs/err.D_OP_INCOMPAT.dupgtype.d
\
- ${TESTSRCDIR}/tst/common/assocs/err.D_OP_INCOMPAT.dupttype.d
\
- ${TESTSRCDIR}/tst/common/assocs/tst.orthogonality.d
\
- ${TESTSRCDIR}/tst/common/buffering/tst.cputime.ksh
\
- ${TESTSRCDIR}/tst/common/buffering/tst.ring3.d
\
- ${TESTSRCDIR}/tst/common/builtinvar/tst.psinfo.d
\
- ${TESTSRCDIR}/tst/common/builtinvar/tst.psinfo1.d
\
- ${TESTSRCDIR}/tst/common/drops/drp.DTRACEDROP_STKSTROVERFLOW.d
\
-
${TESTSRCDIR}/tst/common/dtraceUtil/err.D_PDESC_ZERO.InvalidDescription1.d
\
- ${TESTSRCDIR}/tst/common/dtraceUtil/tst.DataModel32.d.ksh
\
- ${TESTSRCDIR}/tst/common/dtraceUtil/tst.DestructWithFunction.d.ksh
\
- ${TESTSRCDIR}/tst/common/dtraceUtil/tst.DestructWithModule.d.ksh
\
- ${TESTSRCDIR}/tst/common/dtraceUtil/tst.PreprocessorStatement.d.ksh
\
- ${TESTSRCDIR}/tst/common/dtraceUtil/tst.ZeroFunctionProbes.d.ksh
\
- ${TESTSRCDIR}/tst/common/funcs/err.D_PROTO_ARG.mobadarg.d
\
- ${TESTSRCDIR}/tst/common/funcs/err.D_PROTO_LEN.motoofew.d
\
- ${TESTSRCDIR}/tst/common/funcs/err.D_PROTO_LEN.motoomany.d
\
- ${TESTSRCDIR}/tst/common/funcs/err.D_PROTO_LEN.mtabadarg.d
\
- ${TESTSRCDIR}/tst/common/funcs/err.D_PROTO_LEN.mtatoofew.d
\
- ${TESTSRCDIR}/tst/common/funcs/err.D_PROTO_LEN.mtatoomany.d
\
- ${TESTSRCDIR}/tst/common/funcs/tst.bcopy.d
\
- ${TESTSRCDIR}/tst/common/funcs/tst.copyin.d
\
- ${TESTSRCDIR}/tst/common/funcs/tst.copyinto.d
\
- ${TESTSRCDIR}/tst/common/funcs/tst.ddi_pathname.d
\
- ${TESTSRCDIR}/tst/common/inline/err.D_OP_INCOMPAT.badxlate.d
\
- ${TESTSRCDIR}/tst/common/inline/tst.InlineDataAssign.d
\
- ${TESTSRCDIR}/tst/common/inline/tst.InlineExpression.d
\
- ${TESTSRCDIR}/tst/common/inline/tst.InlineTypedef.d
\
- ${TESTSRCDIR}/tst/common/io/tst.fds.c
\
- ${TESTSRCDIR}/tst/common/java_api/tst.FunctionLookup.c
\
- ${TESTSRCDIR}/tst/common/java_api/tst.ProbeData.c
\
- ${TESTSRCDIR}/tst/common/misc/tst.dynopt.d
\
- ${TESTSRCDIR}/tst/common/misc/tst.include.ksh
\
- ${TESTSRCDIR}/tst/common/misc/tst.roch.d
\
- ${TESTSRCDIR}/tst/common/misc/tst.schrock.ksh
\
- ${TESTSRCDIR}/tst/common/nfs/tst.call.c
\
- ${TESTSRCDIR}/tst/common/nfs/tst.call.d
\
- ${TESTSRCDIR}/tst/common/nfs/tst.call3.c
\
- ${TESTSRCDIR}/tst/common/nfs/tst.call3.d
\
- ${TESTSRCDIR}/tst/common/offsetof/err.D_OFFSETOF_BITFIELD.bitfield.d
\
- ${TESTSRCD
Re: svn commit: r211561 - head/cddl/lib/drti
On 21 Aug 2010, at 14:59, Kostik Belousov wrote: > On Sat, Aug 21, 2010 at 01:48:05PM +, Rui Paulo wrote: >> Author: rpaulo >> Date: Sat Aug 21 13:48:04 2010 >> New Revision: 211561 >> URL: http://svn.freebsd.org/changeset/base/211561 >> >> Log: >> Link drti.o with a PIC version of libelf. This is needed because >> drti.o depends on libelf and this avoids linking every other drti.o >> program (namely programs with USDT probes) with libelf. > What is wrong with having libelf as dependency of the resulting > binary ? The resulting binary can be a library. If I add drti.o to the libthr library, every other program will break because it can't find libelf symbols. The only way to fix it would be to rebuild every program that depends on libthr and I don't want that to happen. Regards, -- Rui Paulo ___ 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: r211561 - head/cddl/lib/drti
On Sat, Aug 21, 2010 at 03:16:49PM +0100, Rui Paulo wrote: > > On 21 Aug 2010, at 14:59, Kostik Belousov wrote: > > > On Sat, Aug 21, 2010 at 01:48:05PM +, Rui Paulo wrote: > >> Author: rpaulo > >> Date: Sat Aug 21 13:48:04 2010 > >> New Revision: 211561 > >> URL: http://svn.freebsd.org/changeset/base/211561 > >> > >> Log: > >> Link drti.o with a PIC version of libelf. This is needed because > >> drti.o depends on libelf and this avoids linking every other drti.o > >> program (namely programs with USDT probes) with libelf. > > What is wrong with having libelf as dependency of the resulting > > binary ? > > The resulting binary can be a library. If I add drti.o to the libthr > library, every other program will break because it can't find libelf > symbols. The only way to fix it would be to rebuild every program that > depends on libthr and I don't want that to happen. I do not understand this. If drti.o is added to a library, library should also record a dependency on libelf, i.e. be linked with it. Since you already modifying link command to add drti.o, what is the problem to add -lelf ? pgpwwIDmE2SLM.pgp Description: PGP signature
svn commit: r211568 - head/sys/sparc64/sparc64
Author: marius
Date: Sat Aug 21 14:28:48 2010
New Revision: 211568
URL: http://svn.freebsd.org/changeset/base/211568
Log:
Skip a KASSERT which isn't appropriate when not employing page coloring.
Reported by: Michael Moll
Modified:
head/sys/sparc64/sparc64/pmap.c
Modified: head/sys/sparc64/sparc64/pmap.c
==
--- head/sys/sparc64/sparc64/pmap.c Sat Aug 21 14:14:24 2010
(r211567)
+++ head/sys/sparc64/sparc64/pmap.c Sat Aug 21 14:28:48 2010
(r211568)
@@ -833,14 +833,15 @@ pmap_cache_remove(vm_page_t m, vm_offset
m->md.colors[DCACHE_COLOR(va)]);
KASSERT((m->flags & PG_FICTITIOUS) == 0,
("pmap_cache_remove: fake page"));
- KASSERT(m->md.colors[DCACHE_COLOR(va)] > 0,
- ("pmap_cache_remove: no mappings %d <= 0",
- m->md.colors[DCACHE_COLOR(va)]));
PMAP_STATS_INC(pmap_ncache_remove);
if (dcache_color_ignore != 0)
return;
+ KASSERT(m->md.colors[DCACHE_COLOR(va)] > 0,
+ ("pmap_cache_remove: no mappings %d <= 0",
+ m->md.colors[DCACHE_COLOR(va)]));
+
/*
* Find the color for this virtual address and note the removal of
* the mapping.
___
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: r211561 - head/cddl/lib/drti
On 21 Aug 2010, at 15:23, Kostik Belousov wrote: > On Sat, Aug 21, 2010 at 03:16:49PM +0100, Rui Paulo wrote: >> >> On 21 Aug 2010, at 14:59, Kostik Belousov wrote: >> >>> On Sat, Aug 21, 2010 at 01:48:05PM +, Rui Paulo wrote: Author: rpaulo Date: Sat Aug 21 13:48:04 2010 New Revision: 211561 URL: http://svn.freebsd.org/changeset/base/211561 Log: Link drti.o with a PIC version of libelf. This is needed because drti.o depends on libelf and this avoids linking every other drti.o program (namely programs with USDT probes) with libelf. >>> What is wrong with having libelf as dependency of the resulting >>> binary ? >> >> The resulting binary can be a library. If I add drti.o to the libthr >> library, every other program will break because it can't find libelf >> symbols. The only way to fix it would be to rebuild every program that >> depends on libthr and I don't want that to happen. > > I do not understand this. If drti.o is added to a library, library should > also record a dependency on libelf, i.e. be linked with it. Since you > already modifying link command to add drti.o, what is the problem to > add -lelf ? I wanted the same steps to be done in FreeBSD as in Solaris. I.e., in Solaris you don't need to link with -lelf your program after you added DTrace probes to it. I believe this is a good tradeoff. Regards, -- Rui Paulo ___ 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: r211570 - head/gnu/lib/libobjc
Author: rpaulo
Date: Sat Aug 21 14:59:37 2010
New Revision: 211570
URL: http://svn.freebsd.org/changeset/base/211570
Log:
We need to pull bsd.own.mk before modifying the CC variable. Otherwise
it will be overwritten when we include bsd.lib.mk.
Pointed out by: Dimitry Andric
Modified:
head/gnu/lib/libobjc/Makefile
Modified: head/gnu/lib/libobjc/Makefile
==
--- head/gnu/lib/libobjc/Makefile Sat Aug 21 14:50:16 2010
(r211569)
+++ head/gnu/lib/libobjc/Makefile Sat Aug 21 14:59:37 2010
(r211570)
@@ -1,5 +1,7 @@
# $FreeBSD$
+.include
+
OBJCDIR=${.CURDIR}/../../../contrib/libobjc
GCCDIR=${.CURDIR}/../../../contrib/gcc
GCCLIB=${.CURDIR}/../../../contrib/gcclibs
___
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: r211571 - head/sys/boot/i386/boot2
Author: rpaulo
Date: Sat Aug 21 15:01:59 2010
New Revision: 211571
URL: http://svn.freebsd.org/changeset/base/211571
Log:
Make sure the boot2 stage is compiled with gcc, as clang has no
problems compiling it, but it just gets too big at the moment, even
with -Os. This is not applicable to gptboot, though.
Submitted by: Dimitry Andric
Modified:
head/sys/boot/i386/boot2/Makefile
Modified: head/sys/boot/i386/boot2/Makefile
==
--- head/sys/boot/i386/boot2/Makefile Sat Aug 21 14:59:37 2010
(r211570)
+++ head/sys/boot/i386/boot2/Makefile Sat Aug 21 15:01:59 2010
(r211571)
@@ -1,5 +1,10 @@
# $FreeBSD$
+.include
+
+# XXX: clang can compile the boot code just fine, but boot2 gets too big
+CC:=${CC:C/^cc|^clang/gcc/}
+
FILES= boot boot1 boot2
NM?= nm
___
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: r211572 - head
Author: rpaulo
Date: Sat Aug 21 15:39:06 2010
New Revision: 211572
URL: http://svn.freebsd.org/changeset/base/211572
Log:
Add clang to the cross-tools build stage.
Submitted by: Dimitry Andric
Modified:
head/Makefile.inc1
Modified: head/Makefile.inc1
==
--- head/Makefile.inc1 Sat Aug 21 15:01:59 2010(r211571)
+++ head/Makefile.inc1 Sat Aug 21 15:39:06 2010(r211572)
@@ -1078,8 +1078,15 @@ _kgzip= usr.sbin/kgzip
.endif
.endif
+.if ${CC:T:Mclang} == "clang"
+_clang=usr.bin/clang
+_clang_libs= lib/clang
+.endif
+
cross-tools:
.for _tool in \
+${_clang_libs} \
+${_clang} \
gnu/usr.bin/binutils \
gnu/usr.bin/cc \
usr.bin/xlint/lint1 usr.bin/xlint/lint2 usr.bin/xlint/xlint \
___
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: r211573 - in head: contrib/llvm/tools/clang/lib/Driver contrib/llvm/tools/clang/lib/Frontend lib/clang
Author: rpaulo
Date: Sat Aug 21 15:46:33 2010
New Revision: 211573
URL: http://svn.freebsd.org/changeset/base/211573
Log:
Modify clang so that when TOOLS_PREFIX is defined we register the
CLANG_PREFIX macro. This changes the default header search path when we
are building clang as part of cross-tools.
Submitted by: Dimitry Andric
Reviewed by: freebsd-current
Modified:
head/contrib/llvm/tools/clang/lib/Driver/ToolChains.cpp
head/contrib/llvm/tools/clang/lib/Frontend/InitHeaderSearch.cpp
head/lib/clang/clang.build.mk
Modified: head/contrib/llvm/tools/clang/lib/Driver/ToolChains.cpp
==
--- head/contrib/llvm/tools/clang/lib/Driver/ToolChains.cpp Sat Aug 21
15:39:06 2010(r211572)
+++ head/contrib/llvm/tools/clang/lib/Driver/ToolChains.cpp Sat Aug 21
15:46:33 2010(r211573)
@@ -26,6 +26,10 @@
#include // ::getenv
+#ifndef CLANG_PREFIX
+#define CLANG_PREFIX
+#endif
+
using namespace clang::driver;
using namespace clang::driver::toolchains;
@@ -869,11 +873,9 @@ FreeBSD::FreeBSD(const HostInfo &Host, c
getProgramPaths().push_back(getDriver().Dir + "/../libexec");
getProgramPaths().push_back("/usr/libexec");
if (Lib32) {
-getFilePaths().push_back(getDriver().Dir + "/../lib32");
-getFilePaths().push_back("/usr/lib32");
+getFilePaths().push_back(CLANG_PREFIX "/usr/lib32");
} else {
-getFilePaths().push_back(getDriver().Dir + "/../lib");
-getFilePaths().push_back("/usr/lib");
+getFilePaths().push_back(CLANG_PREFIX "/usr/lib");
}
}
Modified: head/contrib/llvm/tools/clang/lib/Frontend/InitHeaderSearch.cpp
==
--- head/contrib/llvm/tools/clang/lib/Frontend/InitHeaderSearch.cpp Sat Aug
21 15:39:06 2010(r211572)
+++ head/contrib/llvm/tools/clang/lib/Frontend/InitHeaderSearch.cpp Sat Aug
21 15:46:33 2010(r211573)
@@ -30,6 +30,9 @@
#define WIN32_LEAN_AND_MEAN 1
#include
#endif
+#ifndef CLANG_PREFIX
+#define CLANG_PREFIX
+#endif
using namespace clang;
using namespace clang::frontend;
@@ -408,9 +411,10 @@ static bool getWindowsSDKDir(std::string
void InitHeaderSearch::AddDefaultCIncludePaths(const llvm::Triple &triple,
const HeaderSearchOptions &HSOpts)
{
-#if 0 /* Remove unneeded include paths. */
// FIXME: temporary hack: hard-coded paths.
- AddPath("/usr/local/include", System, true, false, false);
+#ifndef __FreeBSD__
+ AddPath(CLANG_PREFIX "/usr/local/include", System, true, false, false);
+#endif
// Builtin includes use #include_next directives and should be positioned
// just prior C include dirs.
@@ -421,7 +425,6 @@ void InitHeaderSearch::AddDefaultCInclud
P.appendComponent("include");
AddPath(P.str(), System, false, false, false, /*IgnoreSysRoot=*/ true);
}
-#endif
// Add dirs specified via 'configure --with-c-include-dirs'.
llvm::StringRef CIncludeDirs(C_INCLUDE_DIRS);
@@ -518,13 +521,15 @@ void InitHeaderSearch::AddDefaultCInclud
case llvm::Triple::MinGW32:
AddPath("c:/mingw/include", System, true, false, false);
break;
+ case llvm::Triple::FreeBSD:
+AddPath(CLANG_PREFIX "/usr/include/clang/" CLANG_VERSION_STRING,
+ System, false, false, false);
+break;
default:
break;
}
- AddPath("/usr/include/clang/" CLANG_VERSION_STRING,
-System, false, false, false);
- AddPath("/usr/include", System, false, false, false);
+ AddPath(CLANG_PREFIX "/usr/include", System, false, false, false);
}
void InitHeaderSearch::
@@ -726,8 +731,10 @@ AddDefaultCPlusPlusIncludePaths(const ll
case llvm::Triple::FreeBSD:
// FreeBSD 8.0
// FreeBSD 7.3
-AddGnuCPlusPlusIncludePaths("/usr/include/c++/4.2", "", "", "", triple);
-AddGnuCPlusPlusIncludePaths("/usr/include/c++/4.2/backward", "", "", "",
triple);
+AddGnuCPlusPlusIncludePaths(CLANG_PREFIX "/usr/include/c++/4.2",
+"", "", "", triple);
+AddGnuCPlusPlusIncludePaths(CLANG_PREFIX "/usr/include/c++/4.2/backward",
+"", "", "", triple);
break;
case llvm::Triple::Minix:
AddGnuCPlusPlusIncludePaths("/usr/gnu/include/c++/4.4.3",
Modified: head/lib/clang/clang.build.mk
==
--- head/lib/clang/clang.build.mk Sat Aug 21 15:39:06 2010
(r211572)
+++ head/lib/clang/clang.build.mk Sat Aug 21 15:46:33 2010
(r211573)
@@ -28,6 +28,10 @@ LLVM_REQUIRES_RTTI=
CFLAGS+=-fno-rtti
.endif
+.ifdef TOOLS_PREFIX
+CFLAGS+=-DCLANG_PREFIX=\"${TOOLS_PREFIX}\"
+.endif
+
.PATH: ${LLVM_SRCS}/${SRCDIR}
TBLGEN=tblgen ${CFLAGS:M-I*}
___
svn-src-head@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-head
To unsubscribe, send any mail
Re: svn commit: r211530 - head/sys/netinet6
Hi,
> On Fri, 20 Aug 2010 22:19:37 +0400
> pluknet said:
pluknet> I think it still continues to test the remain condition
pluknet> regardless of test result of the first part (!optp).
No, I don't think my fix is wrong. If the evaluation matches, the
latter should not be evaluated.
pluknet> That should work:
pluknet> - if (!optp->ip6po_pktinfo ||
!optp->ip6po_pktinfo->ipi6_ifindex)
pluknet> + if (optp != NULL && (!optp->ip6po_pktinfo ||
pluknet> + !optp->ip6po_pktinfo->ipi6_ifindex))
With your proposal, when optp is NULL, use_defzone doesn't set. It is
not desired thing.
Sincerely,
--
Hajimu UMEMOTO @ Internet Mutual Aid Society Yokohama, Japan
u...@mahoroba.org u...@{,jp.}FreeBSD.org
http://www.imasy.org/~ume/
___
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: r211561 - head/cddl/lib/drti
On Sat, Aug 21, 2010 at 03:33:35PM +0100, Rui Paulo wrote: > > On 21 Aug 2010, at 15:23, Kostik Belousov wrote: > > > On Sat, Aug 21, 2010 at 03:16:49PM +0100, Rui Paulo wrote: > >> > >> On 21 Aug 2010, at 14:59, Kostik Belousov wrote: > >> > >>> On Sat, Aug 21, 2010 at 01:48:05PM +, Rui Paulo wrote: > Author: rpaulo > Date: Sat Aug 21 13:48:04 2010 > New Revision: 211561 > URL: http://svn.freebsd.org/changeset/base/211561 > > Log: > Link drti.o with a PIC version of libelf. This is needed because > drti.o depends on libelf and this avoids linking every other drti.o > program (namely programs with USDT probes) with libelf. > >>> What is wrong with having libelf as dependency of the resulting > >>> binary ? > >> > >> The resulting binary can be a library. If I add drti.o to the libthr > >> library, every other program will break because it can't find libelf > >> symbols. The only way to fix it would be to rebuild every program that > >> depends on libthr and I don't want that to happen. > > > > I do not understand this. If drti.o is added to a library, library should > > also record a dependency on libelf, i.e. be linked with it. Since you > > already modifying link command to add drti.o, what is the problem to > > add -lelf ? > > I wanted the same steps to be done in FreeBSD as in Solaris. I.e., in > I Solaris you don't need to link with -lelf your program after you > I added DTrace probes to it believe this is a good tradeoff . The drawbacks of linking library statically are, at least: - Wasted space. - Harder upgrade when libelf changes. Much harder for ports and hand- compiled programs. - The mess that happen when the binary also linked with -lelf, and either your drti.o is linked, or a library that linked with its dcrti.o. Pro argument seems to be that one less diff with Solaris code is needed. This is becoming rather long. pgpIFggeasjyW.pgp Description: PGP signature
Re: svn commit: r211561 - head/cddl/lib/drti
On 21 Aug 2010, at 18:06, Kostik Belousov wrote: > On Sat, Aug 21, 2010 at 03:33:35PM +0100, Rui Paulo wrote: >> >> On 21 Aug 2010, at 15:23, Kostik Belousov wrote: >> >>> On Sat, Aug 21, 2010 at 03:16:49PM +0100, Rui Paulo wrote: On 21 Aug 2010, at 14:59, Kostik Belousov wrote: > On Sat, Aug 21, 2010 at 01:48:05PM +, Rui Paulo wrote: >> Author: rpaulo >> Date: Sat Aug 21 13:48:04 2010 >> New Revision: 211561 >> URL: http://svn.freebsd.org/changeset/base/211561 >> >> Log: >> Link drti.o with a PIC version of libelf. This is needed because >> drti.o depends on libelf and this avoids linking every other drti.o >> program (namely programs with USDT probes) with libelf. > What is wrong with having libelf as dependency of the resulting > binary ? The resulting binary can be a library. If I add drti.o to the libthr library, every other program will break because it can't find libelf symbols. The only way to fix it would be to rebuild every program that depends on libthr and I don't want that to happen. >>> >>> I do not understand this. If drti.o is added to a library, library should >>> also record a dependency on libelf, i.e. be linked with it. Since you >>> already modifying link command to add drti.o, what is the problem to >>> add -lelf ? >> >> I wanted the same steps to be done in FreeBSD as in Solaris. I.e., in >> I Solaris you don't need to link with -lelf your program after you >> I added DTrace probes to it believe this is a good tradeoff . > > The drawbacks of linking library statically are, at least: > - Wasted space. This is certainly not a problem nowadays. > - Harder upgrade when libelf changes. Much harder for ports and hand- > compiled programs. > - The mess that happen when the binary also linked with -lelf, and > either your drti.o is linked, or a library that linked with its > dcrti.o. I'll try to find a way to make DTrace do the linking, if possible. Regards, -- Rui Paulo ___ 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: r211577 - in head: . cddl/lib/drti lib/libelf
Author: rpaulo
Date: Sat Aug 21 17:29:59 2010
New Revision: 211577
URL: http://svn.freebsd.org/changeset/base/211577
Log:
Don't link drti.o with libelf_pic.a. This means that all software with
userland SDT probes must be linked with libelf explicitly.
Requested by: kib
Sponsored by: The FreeBSD Foundation
Modified:
head/Makefile.inc1
head/cddl/lib/drti/Makefile
head/lib/libelf/Makefile
Modified: head/Makefile.inc1
==
--- head/Makefile.inc1 Sat Aug 21 17:15:37 2010(r211576)
+++ head/Makefile.inc1 Sat Aug 21 17:29:59 2010(r211577)
@@ -1149,7 +1149,7 @@ gnu/lib/libgcc__L: lib/libc__L
_prebuild_libs=${_kerberos5_lib_libasn1} ${_kerberos5_lib_libheimntlm}
\
${_kerberos5_lib_libhx509} ${_kerberos5_lib_libkrb5} \
${_kerberos5_lib_libroken} \
- lib/libbz2 lib/libcom_err lib/libcrypt lib/libelf \
+ lib/libbz2 lib/libcom_err lib/libcrypt \
lib/libexpat lib/libfetch \
${_lib_libgssapi} ${_lib_libipx} \
lib/libkiconv lib/libkvm lib/liblzma lib/libmd \
Modified: head/cddl/lib/drti/Makefile
==
--- head/cddl/lib/drti/Makefile Sat Aug 21 17:15:37 2010(r211576)
+++ head/cddl/lib/drti/Makefile Sat Aug 21 17:29:59 2010(r211577)
@@ -8,7 +8,7 @@ FILESOWN= ${LIBOWN}
FILESGRP= ${LIBGRP}
FILESMODE= ${LIBMODE}
FILESDIR= ${LIBDIR}/dtrace
-CLEANFILES=${FILES} ${FILES}.tmp
+CLEANFILES=${FILES}
CFLAGS+= -I${.CURDIR}/../../../sys/cddl/compat/opensolaris \
-I${.CURDIR}/../../../cddl/compat/opensolaris/include \
@@ -18,10 +18,4 @@ CFLAGS+= -I${.CURDIR}/../../../sys/cddl/
-I${OPENSOLARIS_SYS_DISTDIR}/uts/common \
-DPIC -fpic
-.c.o:
- ${CC} ${CFLAGS} -c ${.IMPSRC}
- mv ${FILES} ${FILES}.tmp
- ${LD} -o ${FILES} -r ${FILES}.tmp ${LIBELF_PIC}
- rm -f ${FILES}.tmp
-
.include
Modified: head/lib/libelf/Makefile
==
--- head/lib/libelf/MakefileSat Aug 21 17:15:37 2010(r211576)
+++ head/lib/libelf/MakefileSat Aug 21 17:29:59 2010(r211577)
@@ -55,7 +55,6 @@ SRCS= elf_begin.c
\
${GENSRCS}
INCS= libelf.h gelf.h
-INSTALL_PIC_ARCHIVE=yes
GENSRCS= libelf_fsize.c libelf_msize.c libelf_convert.c
CLEANFILES=${GENSRCS}
CFLAGS+= -I. -I${.CURDIR}
___
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: r211577 - in head: . cddl/lib/drti lib/libelf
On Sat, Aug 21, 2010 at 05:29:59PM +, Rui Paulo wrote: > Author: rpaulo > Date: Sat Aug 21 17:29:59 2010 > New Revision: 211577 > URL: http://svn.freebsd.org/changeset/base/211577 > > Log: > Don't link drti.o with libelf_pic.a. This means that all software with > userland SDT probes must be linked with libelf explicitly. > > Requested by: kib Thanks. pgpfJQKj2pe59.pgp Description: PGP signature
Re: svn commit: r211577 - in head: . cddl/lib/drti lib/libelf
Rui Paulo writes:
> Author: rpaulo
> Date: Sat Aug 21 17:29:59 2010
> New Revision: 211577
> URL: http://svn.freebsd.org/changeset/base/211577
>
> Log:
> Don't link drti.o with libelf_pic.a. This means that all software with
> userland SDT probes must be linked with libelf explicitly.
>
> Requested by: kib
> Sponsored by: The FreeBSD Foundation
>
> Modified:
> head/Makefile.inc1
> head/cddl/lib/drti/Makefile
> head/lib/libelf/Makefile
>
I think there was one more file.
> Author: rpaulo
> Date: Sat Aug 21 14:08:04 2010
> New Revision: 211565
> URL: http://svn.freebsd.org/changeset/base/211565
>
> Log:
> Add the definition of LIBELF_PIC.
>
> Sponsored by: The FreeBSD Foundation
>
> Modified:
> head/share/mk/bsd.libnames.mk
>
> Modified: head/share/mk/bsd.libnames.mk
> ==
> --- head/share/mk/bsd.libnames.mk Sat Aug 21 14:02:04 2010
> (r211564)
> +++ head/share/mk/bsd.libnames.mk Sat Aug 21 14:08:04 2010
> (r211565)
> @@ -45,6 +45,7 @@ LIBDTRACE?= ${DESTDIR}${LIBDIR}/libdtrac
> LIBDWARF?= ${DESTDIR}${LIBDIR}/libdwarf.a
> LIBEDIT?=${DESTDIR}${LIBDIR}/libedit.a
> LIBELF?= ${DESTDIR}${LIBDIR}/libelf.a
> +LIBELF_PIC?= ${DESTDIR}${LIBDIR}/libelf_pic.a
> LIBFETCH?= ${DESTDIR}${LIBDIR}/libfetch.a
> LIBFL?= "don't use LIBFL, use LIBL"
> LIBFORM?=${DESTDIR}${LIBDIR}/libform.a
___
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: r211580 - head/share/mk
Author: rpaulo
Date: Sat Aug 21 17:53:16 2010
New Revision: 211580
URL: http://svn.freebsd.org/changeset/base/211580
Log:
Remove LIBELF_PIC.
Sponsored by: The FreeBSD Foundation
Modified:
head/share/mk/bsd.libnames.mk
Modified: head/share/mk/bsd.libnames.mk
==
--- head/share/mk/bsd.libnames.mk Sat Aug 21 17:52:33 2010
(r211579)
+++ head/share/mk/bsd.libnames.mk Sat Aug 21 17:53:16 2010
(r211580)
@@ -45,7 +45,6 @@ LIBDTRACE?= ${DESTDIR}${LIBDIR}/libdtrac
LIBDWARF?= ${DESTDIR}${LIBDIR}/libdwarf.a
LIBEDIT?= ${DESTDIR}${LIBDIR}/libedit.a
LIBELF?= ${DESTDIR}${LIBDIR}/libelf.a
-LIBELF_PIC?= ${DESTDIR}${LIBDIR}/libelf_pic.a
LIBFETCH?= ${DESTDIR}${LIBDIR}/libfetch.a
LIBFL?="don't use LIBFL, use LIBL"
LIBFORM?= ${DESTDIR}${LIBDIR}/libform.a
___
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: r211577 - in head: . cddl/lib/drti lib/libelf
On 21 Aug 2010, at 18:36, Anonymous wrote: > Rui Paulo writes: > >> Author: rpaulo >> Date: Sat Aug 21 17:29:59 2010 >> New Revision: 211577 >> URL: http://svn.freebsd.org/changeset/base/211577 >> >> Log: >> Don't link drti.o with libelf_pic.a. This means that all software with >> userland SDT probes must be linked with libelf explicitly. >> >> Requested by: kib >> Sponsored by: The FreeBSD Foundation >> >> Modified: >> head/Makefile.inc1 >> head/cddl/lib/drti/Makefile >> head/lib/libelf/Makefile >> > > I think there was one more file. Thanks, fixed. Regards, -- Rui Paulo ___ 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: r211501 - in head/sys: modules modules/send netinet netinet6 sys
On 08/21/2010 05:07, Rui Paulo wrote: Meanwhile, the fact that in the past code has been committed without man pages is the exact cause of the problem you're describing. It is certainly not a justification for continuing to do it wrong. Reverting code just because it doesn't have a man page is pointless. Everyone can write the damn man page, but not everyone can write code. You should really take this into consideration when proposing a commit to be reverted. So let's assume that having lots of code in the tree without a man page is a problem. Further, let's assume that we don't want the problem to get worse. How do you propose that we accomplish that? Doug -- Improve the effectiveness of your Internet presence with a domain name makeover!http://SupersetSolutions.com/ Computers are useless. They can only give you answers. -- Pablo Picasso ___ 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: r211501 - in head/sys: modules modules/send netinet netinet6 sys
On 21 Aug 2010, at 19:20, Doug Barton wrote: > On 08/21/2010 05:07, Rui Paulo wrote: >>> Meanwhile, the fact that in the past code has been committed >>> without man pages is the exact cause of the problem you're >>> describing. It is certainly not a justification for continuing to >>> do it wrong. >> >> Reverting code just because it doesn't have a man page is pointless. >> Everyone can write the damn man page, but not everyone can write >> code. You should really take this into consideration when proposing a >> commit to be reverted. > > So let's assume that having lots of code in the tree without a man page is a > problem. Further, let's assume that we don't want the problem to get worse. > How do you propose that we accomplish that? By writing man pages...? Like I said, everyone can write a man page.. I did write man pages for stuff that hasn't had a man page for years (ctfmerge, ctfconvert, etc.). Regards, -- Rui Paulo ___ 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: r211501 - in head/sys: modules modules/send netinet netinet6 sys
On Sat, 21 Aug 2010, Rui Paulo wrote: On 21 Aug 2010, at 19:20, Doug Barton wrote: On 08/21/2010 05:07, Rui Paulo wrote: Meanwhile, the fact that in the past code has been committed without man pages is the exact cause of the problem you're describing. It is certainly not a justification for continuing to do it wrong. Reverting code just because it doesn't have a man page is pointless. Everyone can write the damn man page, but not everyone can write code. You should really take this into consideration when proposing a commit to be reverted. So let's assume that having lots of code in the tree without a man page is a problem. Further, let's assume that we don't want the problem to get worse. How do you propose that we accomplish that? By writing man pages...? But that's the status quo (I.e., hope that people will/eventually write man pages for stuff) and it hasn't worked so far. The situation isn't actually going to be improved until we make a change. A simple (and I think abundantly reasonable) change that we can make today is, "Don't allow new work to be committed without a man page." Meanwhile, there are a lot of benefits to this idea above and beyond the obvious one of having man pages for stuff. The biggest single benefit is that actually writing documentation for things forces you to re-examine assumptions, work through non-obvious and/or non-default code paths, etc. Speaking for myself, it's happened on more than one occasion that as I sit down to document some piece of code I realize that there are either bugs, or at least a better way of doing something. Requiring developers to work through this process before code goes into the tree gets us better code in addition to documentation. Doug -- Improve the effectiveness of your Internet presence with a domain name makeover!http://SupersetSolutions.com/ Computers are useless. They can only give you answers. -- Pablo Picasso ___ 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: r211501 - in head/sys: modules modules/send netinet netinet6 sys
On 21 Aug 2010, at 22:15, Doug Barton wrote: > On Sat, 21 Aug 2010, Rui Paulo wrote: > >> On 21 Aug 2010, at 19:20, Doug Barton wrote: >> >>> On 08/21/2010 05:07, Rui Paulo wrote: > Meanwhile, the fact that in the past code has been committed > without man pages is the exact cause of the problem you're > describing. It is certainly not a justification for continuing to > do it wrong. Reverting code just because it doesn't have a man page is pointless. Everyone can write the damn man page, but not everyone can write code. You should really take this into consideration when proposing a commit to be reverted. >>> >>> So let's assume that having lots of code in the tree without a man page is >>> a problem. Further, let's assume that we don't want the problem to get >>> worse. How do you propose that we accomplish that? >> >> By writing man pages...? > > But that's the status quo (I.e., hope that people will/eventually write man > pages for stuff) and it hasn't worked so far. The situation isn't actually > going to be improved until we make a change. A simple (and I think abundantly > reasonable) change that we can make today is, "Don't allow new work to be > committed without a man page." > > Meanwhile, there are a lot of benefits to this idea above and beyond the > obvious one of having man pages for stuff. The biggest single benefit is that > actually writing documentation for things forces you to re-examine > assumptions, work through non-obvious and/or non-default code paths, etc. > Speaking for myself, it's happened on more than one occasion that as I sit > down to document some piece of code I realize that there are either bugs, or > at least a better way of doing something. Requiring developers to work > through this process before code goes into the tree gets us better code in > addition to documentation. If your concern is to change the status quo, you do understand that you picked the worst possible thread to hijack ? Esp. since the author promised to roll out a man page. Regards, -- Rui Paulo ___ 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: r211501 - in head/sys: modules modules/send netinet netinet6 sys
On Sat, 21 Aug 2010, Rui Paulo wrote: If your concern is to change the status quo, you do understand that you picked the worst possible thread to hijack ? Esp. since the author promised to roll out a man page. I didn't hijack anything, I responded to Ana's message saying that a man page was forthcoming. To me, since Ana intends to write a man page anyway I thought this was an excellent opportunity to highlight the fact that committing the man page along with the code is the preferred course of action. Doug -- Improve the effectiveness of your Internet presence with a domain name makeover!http://SupersetSolutions.com/ Computers are useless. They can only give you answers. -- Pablo Picasso ___ 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: r211594 - head/sys/dev/bge
Author: yongari Date: Sat Aug 21 23:13:16 2010 New Revision: 211594 URL: http://svn.freebsd.org/changeset/base/211594 Log: Load tunable from loader.conf(5) instead of device.hints(5). Modified: head/sys/dev/bge/if_bge.c Modified: head/sys/dev/bge/if_bge.c == --- head/sys/dev/bge/if_bge.c Sat Aug 21 22:23:22 2010(r211593) +++ head/sys/dev/bge/if_bge.c Sat Aug 21 23:13:16 2010(r211594) @@ -4898,6 +4898,8 @@ bge_add_sysctls(struct bge_softc *sc) struct sysctl_ctx_list *ctx; struct sysctl_oid_list *children, *schildren; struct sysctl_oid *tree; + char tn[32]; + int unit; ctx = device_get_sysctl_ctx(sc->bge_dev); children = SYSCTL_CHILDREN(device_get_sysctl_tree(sc->bge_dev)); @@ -4917,6 +4919,7 @@ bge_add_sysctls(struct bge_softc *sc) #endif + unit = device_get_unit(sc->bge_dev); /* * A common design characteristic for many Broadcom client controllers * is that they only support a single outstanding DMA read operation @@ -4929,13 +4932,13 @@ bge_add_sysctls(struct bge_softc *sc) * performance is about 850Mbps. However forcing coalescing mbufs * consumes a lot of CPU cycles, so leave it off by default. */ + sc->bge_forced_collapse = 0; + snprintf(tn, sizeof(tn), "dev.bge.%d.forced_collapse", unit); + TUNABLE_INT_FETCH(tn, &sc->bge_forced_collapse); SYSCTL_ADD_INT(ctx, children, OID_AUTO, "forced_collapse", CTLFLAG_RW, &sc->bge_forced_collapse, 0, "Number of fragmented TX buffers of a frame allowed before " "forced collapsing"); - resource_int_value(device_get_name(sc->bge_dev), - device_get_unit(sc->bge_dev), "forced_collapse", - &sc->bge_forced_collapse); if (BGE_IS_5705_PLUS(sc)) return; ___ 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: r211594 - head/sys/dev/bge
In message: <201008212313.o7lndgph083...@svn.freebsd.org> Pyun YongHyeon writes: : Author: yongari : Date: Sat Aug 21 23:13:16 2010 : New Revision: 211594 : URL: http://svn.freebsd.org/changeset/base/211594 : : Log: : Load tunable from loader.conf(5) instead of device.hints(5). Why the move? Warner : Modified: : head/sys/dev/bge/if_bge.c : : Modified: head/sys/dev/bge/if_bge.c : == : --- head/sys/dev/bge/if_bge.c Sat Aug 21 22:23:22 2010(r211593) : +++ head/sys/dev/bge/if_bge.c Sat Aug 21 23:13:16 2010(r211594) : @@ -4898,6 +4898,8 @@ bge_add_sysctls(struct bge_softc *sc) : struct sysctl_ctx_list *ctx; : struct sysctl_oid_list *children, *schildren; : struct sysctl_oid *tree; : + char tn[32]; : + int unit; : : ctx = device_get_sysctl_ctx(sc->bge_dev); : children = SYSCTL_CHILDREN(device_get_sysctl_tree(sc->bge_dev)); : @@ -4917,6 +4919,7 @@ bge_add_sysctls(struct bge_softc *sc) : : #endif : : + unit = device_get_unit(sc->bge_dev); : /* :* A common design characteristic for many Broadcom client controllers :* is that they only support a single outstanding DMA read operation : @@ -4929,13 +4932,13 @@ bge_add_sysctls(struct bge_softc *sc) :* performance is about 850Mbps. However forcing coalescing mbufs :* consumes a lot of CPU cycles, so leave it off by default. :*/ : + sc->bge_forced_collapse = 0; : + snprintf(tn, sizeof(tn), "dev.bge.%d.forced_collapse", unit); : + TUNABLE_INT_FETCH(tn, &sc->bge_forced_collapse); : SYSCTL_ADD_INT(ctx, children, OID_AUTO, "forced_collapse", : CTLFLAG_RW, &sc->bge_forced_collapse, 0, : "Number of fragmented TX buffers of a frame allowed before " : "forced collapsing"); : - resource_int_value(device_get_name(sc->bge_dev), : - device_get_unit(sc->bge_dev), "forced_collapse", : - &sc->bge_forced_collapse); : : if (BGE_IS_5705_PLUS(sc)) : return; : ___ 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: r211501 - in head/sys: modules modules/send netinet netinet6 sys
In message: Doug Barton writes: : On Sat, 21 Aug 2010, Rui Paulo wrote: : : > If your concern is to change the status quo, you do understand that : > you picked the worst possible thread to hijack ? Esp. since the author : > promised to roll out a man page. : : I didn't hijack anything, I responded to Ana's message saying that a : man page was forthcoming. To me, since Ana intends to write a man page : anyway I thought this was an excellent opportunity to highlight the : fact that committing the man page along with the code is the preferred : course of action. Do we really need to have a bunch of threads about why people should do the things they have promised to do? Warner ___ 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: r211594 - head/sys/dev/bge
On Sat, Aug 21, 2010 at 07:13:34PM -0600, M. Warner Losh wrote: > In message: <201008212313.o7lndgph083...@svn.freebsd.org> > Pyun YongHyeon writes: > : Author: yongari > : Date: Sat Aug 21 23:13:16 2010 > : New Revision: 211594 > : URL: http://svn.freebsd.org/changeset/base/211594 > : > : Log: > : Load tunable from loader.conf(5) instead of device.hints(5). > > Why the move? > Because bge(4) man page didn't mention device.hints. The tunable could be set via sysctl or loader. ___ 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: r211596 - head/sys/dev/bge
Author: yongari
Date: Sun Aug 22 01:39:09 2010
New Revision: 211596
URL: http://svn.freebsd.org/changeset/base/211596
Log:
It seems all Broadcom controllers have a bug that can generate UDP
datagrams with checksum value 0 when TX UDP checksum offloading is
enabled. Generating UDP checksum value 0 is RFC 768 violation.
Even though the probability of generating such UDP datagrams is
low, I don't want to see FreeBSD boxes to inject such datagrams
into network so disable UDP checksum offloading by default. Users
still override this behavior by setting a sysctl variable or loader
tunable, dev.bge.%d.forced_udpcsum.
I have no idea why this issue was not reported so far given that
bge(4) is one of the most commonly used controller on high-end
server class systems. Thanks to andre@ who passed the PR to me.
PR: kern/104826
Modified:
head/sys/dev/bge/if_bge.c
head/sys/dev/bge/if_bgereg.h
Modified: head/sys/dev/bge/if_bge.c
==
--- head/sys/dev/bge/if_bge.c Sun Aug 22 00:04:24 2010(r211595)
+++ head/sys/dev/bge/if_bge.c Sun Aug 22 01:39:09 2010(r211596)
@@ -120,7 +120,7 @@ __FBSDID("$FreeBSD$");
#include
-#defineBGE_CSUM_FEATURES (CSUM_IP | CSUM_TCP | CSUM_UDP)
+#defineBGE_CSUM_FEATURES (CSUM_IP | CSUM_TCP)
#defineETHER_MIN_NOPAD (ETHER_MIN_LEN - ETHER_CRC_LEN) /*
i.e., 60 */
MODULE_DEPEND(bge, pci, 1, 1, 1);
@@ -2795,6 +2795,8 @@ bge_attach(device_t dev)
goto fail;
}
+ bge_add_sysctls(sc);
+
/* Set default tuneable values. */
sc->bge_stat_ticks = BGE_TICKS_PER_SEC;
sc->bge_rx_coal_ticks = 150;
@@ -2802,6 +2804,11 @@ bge_attach(device_t dev)
sc->bge_rx_max_coal_bds = 10;
sc->bge_tx_max_coal_bds = 10;
+ /* Initialize checksum features to use. */
+ sc->bge_csum_features = BGE_CSUM_FEATURES;
+ if (sc->bge_forced_udpcsum != 0)
+ sc->bge_csum_features |= CSUM_UDP;
+
/* Set up ifnet structure */
ifp = sc->bge_ifp = if_alloc(IFT_ETHER);
if (ifp == NULL) {
@@ -2818,7 +2825,7 @@ bge_attach(device_t dev)
ifp->if_snd.ifq_drv_maxlen = BGE_TX_RING_CNT - 1;
IFQ_SET_MAXLEN(&ifp->if_snd, ifp->if_snd.ifq_drv_maxlen);
IFQ_SET_READY(&ifp->if_snd);
- ifp->if_hwassist = BGE_CSUM_FEATURES;
+ ifp->if_hwassist = sc->bge_csum_features;
ifp->if_capabilities = IFCAP_HWCSUM | IFCAP_VLAN_HWTAGGING |
IFCAP_VLAN_MTU;
if ((sc->bge_flags & BGE_FLAG_TSO) != 0) {
@@ -2975,8 +2982,6 @@ again:
device_printf(sc->bge_dev, "couldn't set up irq\n");
}
- bge_add_sysctls(sc);
-
return (0);
fail:
@@ -3960,7 +3965,7 @@ bge_encap(struct bge_softc *sc, struct m
return (ENOBUFS);
csum_flags |= BGE_TXBDFLAG_CPU_PRE_DMA |
BGE_TXBDFLAG_CPU_POST_DMA;
- } else if ((m->m_pkthdr.csum_flags & BGE_CSUM_FEATURES) != 0) {
+ } else if ((m->m_pkthdr.csum_flags & sc->bge_csum_features) != 0) {
if (m->m_pkthdr.csum_flags & CSUM_IP)
csum_flags |= BGE_TXBDFLAG_IP_CSUM;
if (m->m_pkthdr.csum_flags & (CSUM_TCP | CSUM_UDP)) {
@@ -4237,6 +4242,17 @@ bge_init_locked(struct bge_softc *sc)
/* Program VLAN tag stripping. */
bge_setvlan(sc);
+ /* Override UDP checksum offloading. */
+ if (sc->bge_forced_udpcsum == 0)
+ sc->bge_csum_features &= ~CSUM_UDP;
+ else
+ sc->bge_csum_features |= CSUM_UDP;
+ if (ifp->if_capabilities & IFCAP_TXCSUM &&
+ ifp->if_capenable & IFCAP_TXCSUM) {
+ ifp->if_hwassist &= ~(BGE_CSUM_FEATURES | CSUM_UDP);
+ ifp->if_hwassist |= sc->bge_csum_features;
+ }
+
/* Init RX ring. */
if (bge_init_rx_ring_std(sc) != 0) {
device_printf(sc->bge_dev, "no memory for std Rx buffers.\n");
@@ -4562,9 +4578,9 @@ bge_ioctl(struct ifnet *ifp, u_long comm
ifp->if_capenable ^= IFCAP_HWCSUM;
if (IFCAP_HWCSUM & ifp->if_capenable &&
IFCAP_HWCSUM & ifp->if_capabilities)
- ifp->if_hwassist |= BGE_CSUM_FEATURES;
+ ifp->if_hwassist |= sc->bge_csum_features;
else
- ifp->if_hwassist &= ~BGE_CSUM_FEATURES;
+ ifp->if_hwassist &= ~sc->bge_csum_features;
}
if ((mask & IFCAP_TSO4) != 0 &&
@@ -4940,6 +4956,24 @@ bge_add_sysctls(struct bge_softc *sc)
"Number of fragmented TX buffers of a frame allowed before "
"forced collapsing");
+ /*
+* It seems all Broadcom controllers have a bug that can generate UDP
+* datagra
svn commit: r211597 - head/share/man/man4
Author: yongari Date: Sun Aug 22 01:40:59 2010 New Revision: 211597 URL: http://svn.freebsd.org/changeset/base/211597 Log: Document newly introduced tunable dev.bge.%d.forced_udpcsum. While I'm here also mention tunable dev.bge.%d.forced_collapse. Modified: head/share/man/man4/bge.4 Modified: head/share/man/man4/bge.4 == --- head/share/man/man4/bge.4 Sun Aug 22 01:39:09 2010(r211596) +++ head/share/man/man4/bge.4 Sun Aug 22 01:40:59 2010(r211597) @@ -31,7 +31,7 @@ .\" .\" $FreeBSD$ .\" -.Dd October 7, 2009 +.Dd August 21, 2010 .Dt BGE 4 .Os .Sh NAME @@ -219,6 +219,26 @@ Allow the ASF feature for cooperating wi Can cause system lockup problems on a small number of systems. Enabled by default. .El +.Sh SYSCTL VARIABLES +The following variables are available as both +.Xr sysctl 8 +variables and +.Xr loader 8 +tunables: +.Bl -tag -width indent +.It Va dev.bge.%d.forced_collapse +Allow collapsing multiple transmit buffers into a single buffer +to increase transmit performance with the cost of CPU cycles. +The default value is 0 to disable transmit buffer collapsing. +.It Va dev.bge.%d.forced_udpcsum +Enable UDP transmit checksum offloading even if controller can generate +UDP datagrams with checksum value 0. +UDP datagrams with checksum value 0 can confuse receiver host as it means +sender did not compute UDP checksum. +The default value is 0 which disables UDP transmit checksum offloading. +The interface need to be brought down and up again before a change takes +effect. +.El .Sh DIAGNOSTICS .Bl -diag .It "bge%d: couldn't map memory" ___ 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: r211594 - head/sys/dev/bge
In message: <20100822012433.gb1...@michelle.cdnetworks.com> Pyun YongHyeon writes: : On Sat, Aug 21, 2010 at 07:13:34PM -0600, M. Warner Losh wrote: : > In message: <201008212313.o7lndgph083...@svn.freebsd.org> : > Pyun YongHyeon writes: : > : Author: yongari : > : Date: Sat Aug 21 23:13:16 2010 : > : New Revision: 211594 : > : URL: http://svn.freebsd.org/changeset/base/211594 : > : : > : Log: : > : Load tunable from loader.conf(5) instead of device.hints(5). : > : > Why the move? : > : : Because bge(4) man page didn't mention device.hints. The tunable : could be set via sysctl or loader. It's just that our other drivers use the hints mechanism. Of course, you could easily set either of these in either loader.conf or device.hints... They are basically the same thing... Warner ___ 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: r211594 - head/sys/dev/bge
On Sat, Aug 21, 2010 at 08:10:42PM -0600, M. Warner Losh wrote: > In message: <20100822012433.gb1...@michelle.cdnetworks.com> > Pyun YongHyeon writes: > : On Sat, Aug 21, 2010 at 07:13:34PM -0600, M. Warner Losh wrote: > : > In message: <201008212313.o7lndgph083...@svn.freebsd.org> > : > Pyun YongHyeon writes: > : > : Author: yongari > : > : Date: Sat Aug 21 23:13:16 2010 > : > : New Revision: 211594 > : > : URL: http://svn.freebsd.org/changeset/base/211594 > : > : > : > : Log: > : > : Load tunable from loader.conf(5) instead of device.hints(5). > : > > : > Why the move? > : > > : > : Because bge(4) man page didn't mention device.hints. The tunable > : could be set via sysctl or loader. > > It's just that our other drivers use the hints mechanism. > > Of course, you could easily set either of these in either loader.conf > or device.hints... They are basically the same thing... > Yes, but hints requires "hint" string to be present before driver.unit.keyword string which is not the same syntax of sysctl variable. > Warner ___ 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: r211594 - head/sys/dev/bge
In message: <20100822022001.gc1...@michelle.cdnetworks.com> Pyun YongHyeon writes: : On Sat, Aug 21, 2010 at 08:10:42PM -0600, M. Warner Losh wrote: : > In message: <20100822012433.gb1...@michelle.cdnetworks.com> : > Pyun YongHyeon writes: : > : On Sat, Aug 21, 2010 at 07:13:34PM -0600, M. Warner Losh wrote: : > : > In message: <201008212313.o7lndgph083...@svn.freebsd.org> : > : > Pyun YongHyeon writes: : > : > : Author: yongari : > : > : Date: Sat Aug 21 23:13:16 2010 : > : > : New Revision: 211594 : > : > : URL: http://svn.freebsd.org/changeset/base/211594 : > : > : : > : > : Log: : > : > : Load tunable from loader.conf(5) instead of device.hints(5). : > : > : > : > Why the move? : > : > : > : : > : Because bge(4) man page didn't mention device.hints. The tunable : > : could be set via sysctl or loader. : > : > It's just that our other drivers use the hints mechanism. : > : > Of course, you could easily set either of these in either loader.conf : > or device.hints... They are basically the same thing... : > : : Yes, but hints requires "hint" string to be present before : driver.unit.keyword string which is not the same syntax of sysctl : variable. I guess my bigger point is that we should consider migration now that we have a better mechanism... Warner ___ 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: r211598 - head/sys/fs/tmpfs
Author: ed
Date: Sun Aug 22 05:36:06 2010
New Revision: 211598
URL: http://svn.freebsd.org/changeset/base/211598
Log:
Add support for whiteouts on tmpfs.
Right now unionfs only allows filesystems to be mounted on top of
another if it supports whiteouts. Even though I have sent a patch to
daichi@ to let unionfs work without it, we'd better also add support for
whiteouts to tmpfs.
This patch implements .vop_whiteout and makes necessary changes to
lookup() and readdir() to take them into account. We must also make sure
that when adding or removing a file, we honour the componentname's
DOWHITEOUT and ISWHITEOUT, to prevent duplicate filenames.
MFC after:1 month
Modified:
head/sys/fs/tmpfs/tmpfs.h
head/sys/fs/tmpfs/tmpfs_subr.c
head/sys/fs/tmpfs/tmpfs_vnops.c
Modified: head/sys/fs/tmpfs/tmpfs.h
==
--- head/sys/fs/tmpfs/tmpfs.h Sun Aug 22 01:40:59 2010(r211597)
+++ head/sys/fs/tmpfs/tmpfs.h Sun Aug 22 05:36:06 2010(r211598)
@@ -72,7 +72,8 @@ struct tmpfs_dirent {
* td_namelen field must always be used when accessing its value. */
char * td_name;
- /* Pointer to the node this entry refers to. */
+ /* Pointer to the node this entry refers to. In case this field
+* is NULL, the node is a whiteout. */
struct tmpfs_node * td_node;
};
@@ -434,6 +435,8 @@ int tmpfs_dir_getdotdent(struct tmpfs_no
inttmpfs_dir_getdotdotdent(struct tmpfs_node *, struct uio *);
struct tmpfs_dirent * tmpfs_dir_lookupbycookie(struct tmpfs_node *, off_t);
inttmpfs_dir_getdents(struct tmpfs_node *, struct uio *, off_t *);
+inttmpfs_dir_whiteout_add(struct vnode *, struct componentname *);
+void tmpfs_dir_whiteout_remove(struct vnode *, struct componentname *);
inttmpfs_reg_resize(struct vnode *, off_t);
inttmpfs_chflags(struct vnode *, int, struct ucred *, struct thread *);
inttmpfs_chmod(struct vnode *, mode_t, struct ucred *, struct thread *);
Modified: head/sys/fs/tmpfs/tmpfs_subr.c
==
--- head/sys/fs/tmpfs/tmpfs_subr.c Sun Aug 22 01:40:59 2010
(r211597)
+++ head/sys/fs/tmpfs/tmpfs_subr.c Sun Aug 22 05:36:06 2010
(r211598)
@@ -261,7 +261,8 @@ tmpfs_alloc_dirent(struct tmpfs_mount *t
memcpy(nde->td_name, name, len);
nde->td_node = node;
- node->tn_links++;
+ if (node != NULL)
+ node->tn_links++;
*de = nde;
@@ -287,9 +288,10 @@ tmpfs_free_dirent(struct tmpfs_mount *tm
struct tmpfs_node *node;
node = de->td_node;
-
- MPASS(node->tn_links > 0);
- node->tn_links--;
+ if (node != NULL) {
+ MPASS(node->tn_links > 0);
+ node->tn_links--;
+ }
}
free(de->td_name, M_TMPFSNAME);
@@ -518,6 +520,8 @@ tmpfs_alloc_file(struct vnode *dvp, stru
/* Now that all required items are allocated, we can proceed to
* insert the new node into the directory, an operation that
* cannot fail. */
+ if (cnp->cn_flags & ISWHITEOUT)
+ tmpfs_dir_whiteout_remove(dvp, cnp);
tmpfs_dir_attach(dvp, de);
out:
@@ -768,39 +772,44 @@ tmpfs_dir_getdents(struct tmpfs_node *no
/* Create a dirent structure representing the current
* tmpfs_node and fill it. */
- d.d_fileno = de->td_node->tn_id;
- switch (de->td_node->tn_type) {
- case VBLK:
- d.d_type = DT_BLK;
- break;
-
- case VCHR:
- d.d_type = DT_CHR;
- break;
-
- case VDIR:
- d.d_type = DT_DIR;
- break;
-
- case VFIFO:
- d.d_type = DT_FIFO;
- break;
-
- case VLNK:
- d.d_type = DT_LNK;
- break;
-
- case VREG:
- d.d_type = DT_REG;
- break;
-
- case VSOCK:
- d.d_type = DT_SOCK;
- break;
-
- default:
- panic("tmpfs_dir_getdents: type %p %d",
- de->td_node, (int)de->td_node->tn_type);
+ if (de->td_node == NULL) {
+ d.d_fileno = 1;
+ d.d_type = DT_WHT;
+ } else {
+ d.d_fileno = de->td_node->tn_id;
+ switch (de->td_node->tn_type) {
+ case VBLK:
+ d.d_type = DT_BLK;
+ break;
+
+ case VCHR:
+
