#
8448afce |
| 09-Oct-2009 |
Attilio Rao <attilio@FreeBSD.org> |
atomic_cmpset_barr_* was added in order to cope with compilers willing to specify their own version of atomic_cmpset_* which could have been different than the membar version.
Right now, however, Fr
atomic_cmpset_barr_* was added in order to cope with compilers willing to specify their own version of atomic_cmpset_* which could have been different than the membar version.
Right now, however, FreeBSD is bound mostly to GCC-like compilers and it is desired to add new support and compat shim mostly when there is a real necessity, in order to avoid too much compatibility bloats.
In this optic, bring back atomic_cmpset_{acq, rel}_* to be the same as atomic_cmpset_* and unwind the atomic_cmpset_barr_* introduction.
Requested by: jhb Reviewed by: jhb Tested by: Giovanni Trematerra <giovanni dot trematerra at gmail dot com>
show more ...
|
#
d9492a44 |
| 06-Oct-2009 |
Attilio Rao <attilio@FreeBSD.org> |
- All the functions in atomic.h needs to be in "physical" form (like not defined through macros or similar) in order to be later compiled in the kernel and offer this way the support for modules
- All the functions in atomic.h needs to be in "physical" form (like not defined through macros or similar) in order to be later compiled in the kernel and offer this way the support for modules (and compatibility among the UP case and SMP case). Fix this for the newly introduced atomic_cmpset_barr_* cases by defining and specifying a template. Note that the new DEFINE_CMPSET_GEN() template save more typing on amd64 than the current code. [1] - Fix the style for memory barriers on amd64.
[1] Reported by: Paul B. Mahol <onemda at gmail dot com>
show more ...
|
#
86d2e48c |
| 06-Oct-2009 |
Attilio Rao <attilio@FreeBSD.org> |
Per their definition, atomic instructions used in conjuction with memory barriers should also ensure that the compiler doesn't reorder paths where they are used. GCC, however, does that aggressively
Per their definition, atomic instructions used in conjuction with memory barriers should also ensure that the compiler doesn't reorder paths where they are used. GCC, however, does that aggressively, even in presence of volatile operands. The most reliable way GCC offers for avoid instructions reordering is clobbering "memory" even if that is theoretically an heavy-weight operation, flushing the content of all the registers and forcing reload of them (We could rely, however, on gcc DTRT by just understanding the purpose as this is a well-known pattern for many modern operating-systems).
Not all our memory barriers, right now, clobber memory for GCC-like compilers. The most notable cases are IA32 and amd64 where the memory barrier are treacted the same as normal atomic instructions. Fix this by offering the possibility to implement atomic instructions with memory barriers separately from the normal version and implement the GCC-like specific one using memory clobbering. Thanks to Chris Lattner (@apple) for his discussion on llvm specifics.
Reported by: jhb Reviewed by: jhb Tested by: rdivacky, Giovanni Trematerra <giovanni dot trematerra at gmail dot com>
show more ...
|
Revision tags: vendor/openssh/5.3p1, vendor/tzdata/tzdata2009n, vendor/x86emu/4.6, vendor/tzdata/tzdata2009m, vendor/acpica/20090903, vendor/openssl/0.9.8k-dtls-fixes, vendor/pf-sys/4.5.002, vendor/pf/4.5, vendor/tzdata/tzdata2009l, vendor/openbsm/1.1-P-2, vendor/acpica/20090730, vendor/bind9/9.4.3-P3, vendor/bind9/9.6.1-P1, vendor/less/v436, vendor/tzcode/tzcode2009k, vendor/tzdata/tzdata2009k, vendor/openbsm/1.1-P-1, vendor/tcsh/6.17.00, vendor/clang/clang-r74788, vendor/llvm/llvm-r74788, vendor/clang/clang-r74383, vendor/llvm/llvm-r74383, vendor/acpica/20090625, vendor/bind9/9.6.1, vendor/clang/clang-r73984, vendor/llvm/llvm-r73984, vendor/clang/clang-r73954, vendor/llvm/llvm-r73954, vendor/clang/clang-r73879, vendor/llvm/llvm-r73879, vendor/tzdata/tzdata2009j, vendor/opensolaris/20080410b, vendor/clang/clang-r73340, vendor/llvm/llvm-r73340, vendor/tzcode/tzcode2009i, vendor/tzdata/tzdata2009i, vendor/clang/clang-r73070, vendor/llvm/llvm-r73070, vendor/openssl/0.9.8k, vendor/clang/clang-r73021, vendor/llvm/llvm-r73021, vendor/clang/clang-r72995, vendor/llvm/llvm-r72995, vendor/clang/clang-r72805, vendor/llvm/llvm-r72805, vendor/clang/clang-r72770, vendor/llvm/llvm-r72770, vendor/clang/clang-r72732, vendor/llvm/llvm-r72732, vendor/acpica/20090521, vendor/acpica/20070320resync, vendor/bind9/9.6.1rc1, vendor/netcat/4.5, vendor/tzcode/tzcode2009h, vendor/tzdata/tzdata2009h, vendor/ee/1.5.0, vendor/ee/1.4.7, vendor/ee/1.4.6, vendor/ee/1.4.5a, vendor/ee/1.4.4, vendor/ee/1.4.3, vendor/ee/1.4.2, vendor/top/3.8b1, vendor/tzcode/tzcode2009e, vendor/file/5.03, vendor/less/v429, vendor/binutils/2.15-r191844-obrien, vendor/file/5.00, release/7.2.0_cvs, release/7.2.0, vendor/tzdata/tzdata2009g, vendor/openbsm/1.1, vendor/tzdata/tzdata2009f, vendor/tzdata/tzdata2009e, vendor/tzdata/tzdata2009d, vendor/bind9/9.4.3-P2, vendor/libpcap/1.0.0, vendor/tcpdump/4.0.0, vendor/tzdata/tzdata2009c, vendor/openbsm/1.1-BETA-1, vendor/wpa/0.6.8, vendor/openssh/5.2p1, vendor/tzdata/tzdata2009b, vendor/gdtoa/20081205, vendor/tzdata/tzdata2009a, vendor/wpa_supplicant/0.5.11, vendor/wpa_supplicant/0.5.10, vendor/openbsm/1.1-ALPHA-5, vendor/bind9/9.3.6-P1, vendor/bind9/9.4.3-P1, vendor/file/4.26, release/7.1.0_cvs, release/7.1.0, vendor/openbsm/1.1-ALPHA-4, vendor/bind9/9.4.3, vendor/resolver/9.4.3, vendor/pf-sys/4.4, vendor/pf-sys/4.3, vendor/pf-sys/4.2, vendor/pf/4.4, vendor/pf/4.3, vendor/pf/4.2 |
|
#
422dcc24 |
| 06-Dec-2008 |
Konstantin Belousov <kib@FreeBSD.org> |
Restore memory clobber, to cause mb on the compiler level too. Use more sane formatting of the assembler.
Pointed out by: bde
|
#
26401731 |
| 05-Dec-2008 |
Konstantin Belousov <kib@FreeBSD.org> |
Unconditionally use locked addition of zero to tip of the stack for memory barriers on i386. It works as a serialization instruction on all IA32 CPUs.
Alternative solution of using {s,l,}fence requi
Unconditionally use locked addition of zero to tip of the stack for memory barriers on i386. It works as a serialization instruction on all IA32 CPUs.
Alternative solution of using {s,l,}fence requires run-time checking of the presense of the corresponding SSE or SSE2 extensions, and possible boot-time patching of the kernel text.
Suggested by: many
show more ...
|
Revision tags: release/6.4.0_cvs, release/6.4.0 |
|
#
db7f0b97 |
| 22-Nov-2008 |
Kip Macy <kmacy@FreeBSD.org> |
- bump __FreeBSD version to reflect added buf_ring, memory barriers, and ifnet functions
- add memory barriers to <machine/atomic.h> - update drivers to only conditionally define their own
- add
- bump __FreeBSD version to reflect added buf_ring, memory barriers, and ifnet functions
- add memory barriers to <machine/atomic.h> - update drivers to only conditionally define their own
- add lockless producer / consumer ring buffer - remove ring buffer implementation from cxgb and update its callers
- add if_transmit(struct ifnet *ifp, struct mbuf *m) to ifnet to allow drivers to efficiently manage multiple hardware queues (i.e. not serialize all packets through one ifq) - expose if_qflush to allow drivers to flush any driver managed queues
This work was supported by Bitgravity Inc. and Chelsio Inc.
show more ...
|
Revision tags: vendor/openbsm/1.1-ALPHA-2, vendor/ncurses/5.7-20081102, vendor/tzdata/tzdata2008i, vendor/tzdata/tzdata2008h, vendor/tzdata/tzdata2008g, vendor/netcat/4.4, vendor/openssl/0.9.8i, vendor/tzdata/tzdata2008f, vendor/gdtoa/20080831, vendor/bind9/9.3.5-P2, vendor/bind9/9.4.2-P2, vendor/sendmail/8.14.3, vendor/ath/0.10.5.10, vendor/ntp/4.2.4p5, vendor/tzdata/tzdata2008e, vendor/tzdata/tzdata2008d, vendor/tzdata/tzdata2008c, vendor/pf-sys/4.1.001, vendor/openssh/5.1p1, vendor/openssh/5.0p1, vendor/openssh/4.9p1, vendor/openssh/4.7p1, vendor/openssh/4.6p1, vendor/bind9/9.3.5-P1, vendor/bind9/9.3.5, vendor/bind9/9.4.2-P1, vendor/misc-GNU/cpio/2.8, vendor/binutils/2.15.20040523, vendor/opensolaris/20080410, vendor/ncurses/5.6-20080503, vendor/expat/2.0.1, vendor/com_err/1.1.0, vendor/heimdal/1.1.0, vendor/opensolaris/20080410a, vendor/gcc/trunk_20080418-SVN134423, vendor/netcat/4.3, vendor/tzdata/tzdata2008b, vendor/bzip2/1.0.5, vendor/misc-GNU/cvs/1.11-20080310 |
|
#
6eb4157f |
| 16-Mar-2008 |
Pawel Jakub Dawidek <pjd@FreeBSD.org> |
Implement atomic_fetchadd_long() for all architectures and document it.
Reviewed by: attilio, jhb, jeff, kris (as a part of the uidinfo_waitfree.patch)
|
Revision tags: vendor/tcsh/6.15b, vendor/tzdata/tzdata2008a, vendor/bsnmp/syrinx_20080307_bugfix, release/7.0.0_cvs, release/7.0.0, vendor/ncurses/5.6-20080209, vendor/file/4.23-r1.46, vendor/binutils/2.15.cvsrev_1_51, release/6.3.0_cvs, release/6.3.0, vendor/file/4.23, vendor/misc-GNU/cvs/1.11.22, vendor/tzdata/tzdata2007k, vendor/ncurses/5.6-20071222, vendor/openpam/HYDRANGEA, vendor/amd/6.1.5, vendor/tzdata/tzdata2007j, vendor/resolver/9.4.2, vendor/bind9/9.4.2, vendor/less/v416, vendor/less/v415, vendor/libreadline/5.2p3_p7, vendor/sendmail/8.14.2, vendor/tzdata/tzdata2007i, vendor/openbsm/1.0, vendor/one-true-awk/20071023, vendor/ipfilter/4.1.28, vendor/ipfilter-sys/4-1-28, vendor/openssl/0.9.8-20071018, vendor/tcpdump/3.9.8, vendor/libpcap/0.9.8, vendor/tcsh/6.15a, vendor/less/v409, vendor/tzdata/tzdata2007h, vendor/less/v408, vendor/tzdata/tzdata2007g, vendor/gcc/4.2.1-20070718-SVN126787, vendor/bind9/9.4.1p1, vendor/openbsm/1.0-ALPHA-15, vendor/wpa_supplicant/0.5.8, vendor/hostapd/0.5.8, vendor/pf/4.1, vendor/pf-sys/4.1, vendor/less/v406, vendor/misc-GNU/diff/2.8.7, vendor/bind9/9.4.1, vendor/one-true-awk/20070501, vendor/ipfilter/4.1.23, vendor/ipfilter-sys/4-1-23, vendor/less/v403, vendor/resolver/9.4.1, vendor/file/4.21, vendor/file/4.19, vendor/tzdata/tzdata2007f, vendor/gcc/4.2.0-20070514-SVN124707, vendor/openbsm/1.0-ALPHA-14, vendor/libreadline/5.2p2, vendor/sendmail/8.14.1, vendor/binutils/2.15.20050217, vendor/tcpdump/3.9.1a, vendor/tcsh/6.15p1, vendor/bzip2/1.0.4, vendor/netcat/4.1, vendor/acpica/20070320, vendor/openssl/0.9.8-20070315, vendor/openssl/0.9.8e, vendor/tcsh/6.15, vendor/bind9/9.3.4, vendor/ncurses/5.6-20061217, release/6.2.0_cvs, release/6.2.0, vendor/gdtoa/20050923, vendor/libreadline/5.2p1, vendor/libreadline/5.2 |
|
#
0b194ec8 |
| 29-Dec-2006 |
Bruce Evans <bde@FreeBSD.org> |
Fix oops in previous commit.
|
#
f28e1c8f |
| 29-Dec-2006 |
Bruce Evans <bde@FreeBSD.org> |
Fixed some style bugs (mainly assorted errors in comments, and inconsistent spelling of `result').
|
#
6c296ffa |
| 29-Dec-2006 |
Bruce Evans <bde@FreeBSD.org> |
Fixed some style bugs (whitespace only).
|
#
7e4277e5 |
| 29-Dec-2006 |
Bruce Evans <bde@FreeBSD.org> |
Try harder to garbage-collect the "LOCORE" (really asm) version of MPLOCKED. The cleaning in rev.1.25 was supposed to have been undone by rev.1.26, but 1.26 could never have actually affected asm fi
Try harder to garbage-collect the "LOCORE" (really asm) version of MPLOCKED. The cleaning in rev.1.25 was supposed to have been undone by rev.1.26, but 1.26 could never have actually affected asm files since atomic.h is full of C declarations so including it in asm files would just give syntax errors. The asm MPLOCKED is even less needed than when misplaced definitions of it were first removed, and is now unused in any asm file in the src tree except in anachronismns in sys/i386/i386/support.s.
show more ...
|
#
26ab2d1d |
| 27-Dec-2006 |
Bruce Evans <bde@FreeBSD.org> |
Avoid an instruction in atomic_cmpset_{int_long)() in most cases. These functions are used a lot for mutexes, so this reduces the text size of an average kernel by about 0.75%. This wasn't intended
Avoid an instruction in atomic_cmpset_{int_long)() in most cases. These functions are used a lot for mutexes, so this reduces the text size of an average kernel by about 0.75%. This wasn't intended to be a significant optimization, but it somehow increased the maximum number of packets per second that can be transmitted by my bge hardware from 320000 to 460000 (this benchmark is CPU-bound and remarkably sensitive to changes in the text section).
Details: we would prefer to leave the result of the cmpxchg in %al, but cannot tell gcc that it is there, so we have to convert it to an integer register. We converted to %al, then to %[re]ax, but the latter step is usually wasted since gcc usually only wants the condition code and can recover it from %al just as easily as from %[re]ax. Let gcc promote %al in the few cases where this is needed.
Nearby style fixes; - let gcc manage the load of `res', and don't abuse `res' for a copy of `exp' - don't echo `res's name in comments - consistently spell the condition code as 'e' after comparison for equality - don't hard-code %al anywhere except in constraints - for the version that doesn't use cmpxchg, there is no requirement to use %al anywhere, so don't hard-code it in the constraints either.
Style non-fix: - for the versions that use cmpxchg, keep using "a" (was %[re]ax, now %al) for the main output operand, although this is not required. The input and output operands that use the "a" constraint are now decoupled, and this makes things clearer except for the reason that the output register is hard-coded. It is now just a hack to tell gcc that the input "a" has been clobbered without increasing the number of operands.
show more ...
|
Revision tags: vendor/resolver/9.3.4, vendor/resolver/9.3.3, vendor/ath/0.9.20.3, vendor/bind9/9.3.3, vendor/pppd/2.3.11, vendor/openssh/4.5p1, vendor/bind9/9.3.2p2, vendor/resolver/9.3.3RC2, vendor/tzdata/tzdata2006n, vendor/openssl/0.9.8-20060929, vendor/openssl/0.9.8d, vendor/openssh/4.4p1, vendor/openbsm/1.0-ALPHA-12, vendor/openbsm/1.0-ALPHA-11, vendor/bind9/9.3.2p1, vendor/tcpdump/3.9.4, vendor/libpcap/0.9.4, vendor/openbsm/1.0-ALPHA-10, vendor/openbsm/1.0-ALPHA9-kwikfix_backout, vendor/NetBSD/lukemftpd/20060831, vendor/openbsm/1.0-ALPHA9-kwikfix, vendor/gcc/3.4.6-20060825-SVN116475, vendor/openbsm/1.0-ALPHA-9, vendor/gdb/6.1.1, vendor/less/v394, vendor/sendmail/8.13.8, vendor/ipfilter/4.1.13, vendor/ipfilter-sys/4-1-13, vendor/openssl/0.9.8b, vendor/libuwx/BETA10, vendor/openbsm/1.0-ALPHA-7, vendor/OpenBSD/cvs_20060501, vendor/file/4.17a, vendor/file/4.17, vendor/sendmail/8.13.7, vendor/binutils/2.15.NOTHING, vendor/ath/0.9.17.2, vendor/openbsm/1.0-ALPHA-6, release/5.5.0_cvs, release/5.5.0, vendor/netcat/3.9, vendor/resolver/9.3.2, vendor/tzdata/tzdata2006g, release/6.1.0_cvs, release/6.1.0 |
|
#
6f0f8cca |
| 28-Mar-2006 |
Dag-Erling Smørgrav <des@FreeBSD.org> |
Use wrapper macros for atomic pointer operations in order to perform the correct casts. This should probably be merged to other architectures.
|
Revision tags: vendor/libreadline/5.1p1-p4, vendor/libreadline/5.1, vendor/zlib/1.2.3, vendor/openssh/4.3p1, vendor/sendmail/8.13.6, vendor/csup/20060313, vendor/hostapd/0.4.8, vendor/wpa_supplicant/0.4.8, vendor/openbsm/1.0-ALPHA-5, vendor/csup/20060303, vendor/csup/20060302, vendor/bsnmp/1.12, vendor/tcsh/6.14, vendor/openbsm/1.0-ALPHA-4, vendor/ath/0.9.16.16, vendor/openbsm/1.0-ALPHA-3, vendor/openbsm/1.0-ALPHA-2, vendor/openbsm/1.0-ALPHA-1, vendor/bind9/9.3.2-dougb-p1, vendor/ipfilter/4.1.10, vendor/ipfilter-sys/4-1-10, vendor/bind9/9.3.2, vendor/tzdata/tzdata2005r, vendor/tzdata/tzdata2005q, vendor/nve/1.0.0310, release/6.0.0_cvs, release/6.0.0, vendor/acpica/20051021, vendor/acpica/20041119sysinc, vendor/wpa_supplicant/0.3-latest, vendor/groff/1.19.2, vendor/wpa_supplicant/0.3.9, vendor/bsnmp/1.11, vendor/misc-GNU/cpio/2.6 |
|
#
3c2bc2bf |
| 27-Sep-2005 |
John Baldwin <jhb@FreeBSD.org> |
Add a new atomic_fetchadd() primitive that atomically adds a value to a variable and returns the previous value of the variable.
Tested on: i386, alpha, sparc64, arm (cognet) Reviewed by: arch@ Subm
Add a new atomic_fetchadd() primitive that atomically adds a value to a variable and returns the previous value of the variable.
Tested on: i386, alpha, sparc64, arm (cognet) Reviewed by: arch@ Submitted by: cognet (arm) MFC after: 1 week
show more ...
|
#
80d52f16 |
| 15-Sep-2005 |
John Baldwin <jhb@FreeBSD.org> |
Stop using the '+' constraint modifier with inline assembly. The '+' constraint is actually only allowed for register operands. Instead, use separate input and output memory constraints.
Education
Stop using the '+' constraint modifier with inline assembly. The '+' constraint is actually only allowed for register operands. Instead, use separate input and output memory constraints.
Education from: alc Reviewed by: alc Tested on: i386, alpha MFC after: 1 week
show more ...
|
Revision tags: vendor/pf-sys/3.7.001, vendor/openssh/4.2p1, vendor/tzdata/tzdata2005m, vendor/tzdata/tzdata2005l, vendor/bind9/9.3.1p1 |
|
#
122eceef |
| 15-Jul-2005 |
John Baldwin <jhb@FreeBSD.org> |
Convert the atomic_ptr() operations over to operating on uintptr_t variables rather than void * variables. This makes it easier and simpler to get asm constraints and volatile keywords correct.
MFC
Convert the atomic_ptr() operations over to operating on uintptr_t variables rather than void * variables. This makes it easier and simpler to get asm constraints and volatile keywords correct.
MFC after: 3 days Tested on: i386, alpha, sparc64 Compiled on: ia64, powerpc, amd64 Kernel toolchain busted on: arm
show more ...
|
Revision tags: vendor/tcpdump/3.9.1, vendor/libpcap/0.9.1 |
|
#
48281036 |
| 09-Jul-2005 |
John Baldwin <jhb@FreeBSD.org> |
Some cleanups and tweaks to some of the atomic.h files in preparation for further changes and fixes in the future: - Use aliases via macros rather than duplicated inlines wherever possible. - Move al
Some cleanups and tweaks to some of the atomic.h files in preparation for further changes and fixes in the future: - Use aliases via macros rather than duplicated inlines wherever possible. - Move all the aliases to the bottom of these files and the inline functions to the top. - Add various comments. - On alpha, drop atomic_{load_acq,store_rel}_{8,char,16,short}(). - On i386 and amd64, don't duplicate the extern declarations for functions in the two non-inline cases (KLD_MODULE and compiler doesn't do inlines), instead, consolidate those two cases. - Some whitespace fixes.
Approved by: re (scottl)
show more ...
|
Revision tags: vendor/openpam/FIGWORT, vendor/sendmail/8.13.4_1, vendor/hostapd/0.3.9, vendor/sendmail/8.13.4, vendor/OpenBSD/dhclient_openbsd_3_7, vendor/hostapd/0.3.7, vendor/wpa_supplicant/0.3.8, vendor/openssh/4.1p1, vendor/openssh/4.0p1, vendor/gcc/3.4.4-20050518, vendor/tcpdump/3.9.1-096, vendor/libpcap/0.9.1-096, vendor/ngatm/1.2, vendor/bsnmp/1.10, vendor/misc-GNU/texinfo/4.8, vendor/NetBSD/lukemftp/20050516, vendor/tnftp/20050516, vendor/one-true-awk/20050424, vendor/bzip2/1.0.3, vendor/libregex/fedora-glibc-2.3.4-21, release/5.4.0_cvs, release/5.4.0, vendor/NetBSD/regress_resolv_20050504, vendor/pf/3.7, vendor/pf-sys/3.7, vendor/bsnmp/1.9, vendor/ipfilter/4.1.8, vendor/ipfilter-sys/4-1-8, vendor/zlib/1.2.2, vendor/bind9/9.3.1, vendor/nve/1.0.0301 |
|
#
a5f50ef9 |
| 02-Mar-2005 |
Joerg Wunsch <joerg@FreeBSD.org> |
netchild's mega-patch to isolate compiler dependencies into a central place.
This moves the dependency on GCC's and other compiler's features into the central sys/cdefs.h file, while the individual
netchild's mega-patch to isolate compiler dependencies into a central place.
This moves the dependency on GCC's and other compiler's features into the central sys/cdefs.h file, while the individual source files can then refer to #ifdef __COMPILER_FEATURE_FOO where they by now used to refer to #if __GNUC__ > 3.1415 && __BARC__ <= 42.
By now, GCC and ICC (the Intel compiler) have been actively tested on IA32 platforms by netchild. Extension to other compilers is supposed to be possible, of course.
Submitted by: netchild Reviewed by: various developers on arch@, some time ago
show more ...
|
Revision tags: vendor/bsnmp/1.8, vendor/openssl/0.9.7e, vendor/heimdal/0.6.3, vendor/NetBSD/lukemftp/20050219, vendor/tnftp/20050219, vendor/sendmail/8.13.3, vendor/netcat/20050204, vendor/openpam/FETERITA, release/4.11.0_cvs, release/4.11.0, vendor/file/4.12, vendor/ath/0.9.14.9, vendor/tzdata/tzdata2004g, vendor/acpica/20041119 |
|
#
2d68e3fb |
| 16-Nov-2004 |
John Baldwin <jhb@FreeBSD.org> |
Initiate deorbit burn sequence for 80386 support in FreeBSD: Remove 80386 (I386_CPU) support from the kernel.
|
#
90baa95f |
| 12-Nov-2004 |
John Baldwin <jhb@FreeBSD.org> |
Spell _KERNEL correctly so that UP kernels are actually optimized again.
Submitted by: pjd
|
#
bd2ed154 |
| 12-Nov-2004 |
John Baldwin <jhb@FreeBSD.org> |
- Use the SMP style ops for atomic_load/store() in userland so that libraries and binaries will work on both UP and SMP machines. - Remove unnecessary gcc memory barrier from the UP atomic_store()
- Use the SMP style ops for atomic_load/store() in userland so that libraries and binaries will work on both UP and SMP machines. - Remove unnecessary gcc memory barrier from the UP atomic_store() op.
Submitted by: bde
show more ...
|
#
57621b8b |
| 11-Nov-2004 |
John Baldwin <jhb@FreeBSD.org> |
- Place the gcc memory barrier hint in the right place in the 80386 version of atomic_store_rel(). - Use the 80386 versions of atomic_load_acq() and atomic_store_rel() that do not use serializing
- Place the gcc memory barrier hint in the right place in the 80386 version of atomic_store_rel(). - Use the 80386 versions of atomic_load_acq() and atomic_store_rel() that do not use serializing instructions on all UP kernels since a UP machine does need to synchronize with other CPUs. This trims lots of cycles from spin locks on UP kernels among other things.
Benchmarked by: rwatson
show more ...
|
Revision tags: release/5.3.0_cvs, release/5.3.0, vendor/openssh/3.9p1, vendor/libreadline/5.0p1-p5, vendor/libreadline/5.0, vendor/tzdata/tzdata2004e, vendor/gcc/3.4.2-20041014, vendor/ngatm/1.1.1, vendor/libbegemot/1.1.1, vendor/bind9/9.3.0, vendor/bind9/9.3.0rc4, vendor/pf/3.5.001, vendor/NetBSD/lukemftpd/20040809, vendor/isc-dhcp/3.0.1, vendor/openssl/0.9.7-dev, vendor/pf-sys/3.5.004, vendor/pf-sys/3.5.003, vendor/misc-GNU/sort/20040812, vendor/gcc/3.4.2-20040728, vendor/ngatm/1.1, vendor/file/4.10, vendor/bsnmp/1.7, vendor/sendmail/8.13.1, vendor/ntp/4.2.0, vendor/pf-sys/3.5.002, vendor/tcsh/6.13, vendor/ngatm/1.0, vendor/amd/6.0.10p1, vendor/misc-GNU/grep/2.5.1, vendor/libregex/from-grep-2.5.1, vendor/misc-GNU/sort/5.2.1, vendor/acpica/20040527, vendor/zlib/1.2.1, vendor/isc-dhcp/3.0.1rc14, vendor/gdb/marcel_contrib, vendor/ipfilter/3.4.35, vendor/ipfilter-sys/3-4-35, vendor/tzcode/tzcode2004a, vendor/pf-sys/3.5.001, vendor/pf/3.5, vendor/pf-sys/3.5, vendor/altq/20040607, vendor/misc-GNU/cvs/1.11.17, release/4.10.0_cvs, release/4.10.0, vendor/acpica/20040514, vendor/libuwx/BETA7, vendor/pf-sys/3.4.003, vendor/NetBSD/lukemftp/20040426, vendor/tnftp/20040426, vendor/openssh/3.8.1p1, vendor/less/v381, vendor/misc-GNU/cvs/1.11.15, vendor/bsnmp/1.6, vendor/acpica/20040402, vendor/pf-sys/3.4.002, vendor/openssl/0.9.7d-p1, vendor/heimdal/0.6.1, vendor/com_err/0.6.1, vendor/tcpdump/3.8.3, vendor/libpcap/0.8.3, vendor/pf-sys/3.4.001, vendor/acpica/20040311, vendor/openssl/0.9.7d, vendor/openssl/0.9.7c-patch1 |
|
#
a122cca9 |
| 12-Mar-2004 |
Tom Rhodes <trhodes@FreeBSD.org> |
These are changes to allow to use the Intel C/C++ compiler (lang/icc) to build the kernel. It doesn't affect the operation if gcc.
Most of the changes are just adding __INTEL_COMPILER to #ifdef's, a
These are changes to allow to use the Intel C/C++ compiler (lang/icc) to build the kernel. It doesn't affect the operation if gcc.
Most of the changes are just adding __INTEL_COMPILER to #ifdef's, as icc v8 may define __GNUC__ some parts may look strange but are necessary.
Additional changes: - in_cksum.[ch]: * use a generic C version instead of the assembly version in the !gcc case (ASM code breaks with the optimizations icc does) -> no bad checksums with an icc compiled kernel Help from: andre, grehan, das Stolen from: alpha version via ppc version The entire checksum code should IMHO be replaced with the DragonFly version (because it isn't guaranteed future revisions of gcc will include similar optimizations) as in: ---snip--- Revision Changes Path 1.12 +1 -0 src/sys/conf/files.i386 1.4 +142 -558 src/sys/i386/i386/in_cksum.c 1.5 +33 -69 src/sys/i386/include/in_cksum.h 1.5 +2 -0 src/sys/netinet/igmp.c 1.6 +0 -1 src/sys/netinet/in.h 1.6 +2 -0 src/sys/netinet/ip_icmp.c
1.4 +3 -4 src/contrib/ipfilter/ip_compat.h 1.3 +1 -2 src/sbin/natd/icmp.c 1.4 +0 -1 src/sbin/natd/natd.c 1.48 +1 -0 src/sys/conf/files 1.2 +0 -1 src/sys/conf/files.amd64 1.13 +0 -1 src/sys/conf/files.i386 1.5 +0 -1 src/sys/conf/files.pc98 1.7 +1 -1 src/sys/contrib/ipfilter/netinet/fil.c 1.10 +2 -3 src/sys/contrib/ipfilter/netinet/ip_compat.h 1.10 +1 -1 src/sys/contrib/ipfilter/netinet/ip_fil.c 1.7 +1 -1 src/sys/dev/netif/txp/if_txp.c 1.7 +1 -1 src/sys/net/ip_mroute/ip_mroute.c 1.7 +1 -2 src/sys/net/ipfw/ip_fw2.c 1.6 +1 -2 src/sys/netinet/igmp.c 1.4 +158 -116 src/sys/netinet/in_cksum.c 1.6 +1 -1 src/sys/netinet/ip_gre.c 1.7 +1 -2 src/sys/netinet/ip_icmp.c 1.10 +1 -1 src/sys/netinet/ip_input.c 1.10 +1 -2 src/sys/netinet/ip_output.c 1.13 +1 -2 src/sys/netinet/tcp_input.c 1.9 +1 -2 src/sys/netinet/tcp_output.c 1.10 +1 -1 src/sys/netinet/tcp_subr.c 1.10 +1 -1 src/sys/netinet/tcp_syncache.c 1.9 +1 -2 src/sys/netinet/udp_usrreq.c
1.5 +1 -2 src/sys/netinet6/ipsec.c 1.5 +1 -2 src/sys/netproto/ipsec/ipsec.c 1.5 +1 -1 src/sys/netproto/ipsec/ipsec_input.c 1.4 +1 -2 src/sys/netproto/ipsec/ipsec_output.c
and finally remove sys/i386/i386 in_cksum.c sys/i386/include in_cksum.h ---snip--- - endian.h: * DTRT in C++ mode - quad.h: * we don't use gcc v1 anymore, remove support for it Suggested by: bde (long ago) - assym.h: * avoid zero-length arrays (remove dependency on a gcc specific feature) This change changes the contents of the object file, but as it's only used to generate some values for a header, and the generator knows how to handle this, there's no impact in the gcc case. Explained by: bde Submitted by: Marius Strobl <marius@alchemy.franken.de> - aicasm.c: * minor change to teach it about the way icc spells "-nostdinc" Not approved by: gibbs (no reply to my mail) - bump __FreeBSD_version (lang/icc needs to know about the changes)
Incarnations of this patch survive gcc compiles since a loooong time, I use it on my desktop. An icc compiled kernel works since Nov. 2003 (exceptions: snd_* if used as modules), it survives a build of the entire ports collection with icc.
Parts of this commit contains suggestions or submissions from Marius Strobl <marius@alchemy.franken.de>.
Reviewed by: -arch Submitted by: netchild
show more ...
|
Revision tags: vendor/misc-GNU/grep/2.4.2, vendor/acpica/20040220, vendor/pf/3.4, vendor/openssh/3.8p1, vendor/pf-sys/3.4, vendor/SGI/vjs_20020502, vendor/libregex/from-grep-2.4.2, release/5.2.1_cvs, release/5.2.1, vendor/sendmail/8.12.11, vendor/libreadline/4.3p1-p5, vendor/openpam/EELGRASS, vendor/one-true-awk/20040207, vendor/one-true-awk/20030731, vendor/bsnmp/1.5a, vendor/gdtoa/20040118, release/5.2.0_cvs, release/5.2.0, vendor/alsa/1.9, vendor/openssh/3.7.1p2, vendor/acpica/20031203, vendor/ath/0.9.6.3, vendor/ath/0.9.6.1, vendor/bind/8.3.7 |
|
#
6d800f89 |
| 17-Nov-2003 |
Bruce Evans <bde@FreeBSD.org> |
Fixed pedantic syntax errors. Many macros didn't permit a semicolon after their invocation in the !KLD_MODULE case, but a semicolon is provided after all invocations and is required in the KLD_MODUL
Fixed pedantic syntax errors. Many macros didn't permit a semicolon after their invocation in the !KLD_MODULE case, but a semicolon is provided after all invocations and is required in the KLD_MODULE case.
show more ...
|
#
ed9800b4 |
| 17-Nov-2003 |
Bruce Evans <bde@FreeBSD.org> |
Avoid a warning for compiling with `gcc -Wbad-function cast'. (This is the warning that points to the bug in `(char *)malloc(...)' where malloc() is implicitly declared as returning int. We do simi
Avoid a warning for compiling with `gcc -Wbad-function cast'. (This is the warning that points to the bug in `(char *)malloc(...)' where malloc() is implicitly declared as returning int. We do similar things here, but they work because u_int is the same as uintptr_t on i386's.)
show more ...
|