Searched hist:w (Results 1 - 25 of 2995) sorted by relevance
| /src/bin/ed/test/ | ||
| H A D | k.d | 1.1 Wed Apr 14 00:05:05 GMT 1993 alm branches: 1.1.120; added tests for the `k' and `w' commands |
| H A D | k.r | 1.1 Wed Apr 14 00:05:06 GMT 1993 alm branches: 1.1.120; added tests for the `k' and `w' commands |
| H A D | k.t | 1.1 Wed Apr 14 00:05:07 GMT 1993 alm branches: 1.1.120; added tests for the `k' and `w' commands |
| H A D | k4.err | 1.1 Wed Apr 14 00:05:08 GMT 1993 alm branches: 1.1.120; added tests for the `k' and `w' commands |
| H A D | w1.err | 1.1 Wed Apr 14 00:05:10 GMT 1993 alm branches: 1.1.120; added tests for the `k' and `w' commands |
| H A D | w2.err | 1.1 Wed Apr 14 00:05:11 GMT 1993 alm branches: 1.1.120; added tests for the `k' and `w' commands |
| H A D | w3.err | 1.1 Wed Apr 14 00:05:12 GMT 1993 alm branches: 1.1.120; added tests for the `k' and `w' commands |
| /src/sys/arch/hpcarm/conf/ | ||
| H A D | std.pxa2x0 | 1.1 Sat Apr 17 13:36:21 GMT 2010 nonaka branches: 1.1.2; 1.1.4; 1.1.6; Added Sharp W-ZERO3 series support. |
| H A D | std.sa11x0 | 1.1 Sat Apr 17 13:36:21 GMT 2010 nonaka branches: 1.1.2; 1.1.4; 1.1.6; Added Sharp W-ZERO3 series support. |
| /src/sys/dev/marvell/ | ||
| H A D | mvxpbmvar.h | 1.1 Wed Jun 03 03:55:47 GMT 2015 hsuenaga branches: 1.1.2; 1.1.18; separate buffer management codes 'mvxpbm.c' from if_mvxpe.c. the buffer management(ex. fill the rx descriptors/buffers) is done by H/W in ARMADA XP/380, and is done by S/W in ARMADA 370. the H/W BM support is not yet implemented, so all devices use the S/W management mode at this time. 1.1 Wed Jun 03 03:55:47 GMT 2015 hsuenaga branches: 1.1.2; 1.1.18; separate buffer management codes 'mvxpbm.c' from if_mvxpe.c. the buffer management(ex. fill the rx descriptors/buffers) is done by H/W in ARMADA XP/380, and is done by S/W in ARMADA 370. the H/W BM support is not yet implemented, so all devices use the S/W management mode at this time. 1.1 Wed Jun 03 03:55:47 GMT 2015 hsuenaga branches: 1.1.2; 1.1.18; separate buffer management codes 'mvxpbm.c' from if_mvxpe.c. the buffer management(ex. fill the rx descriptors/buffers) is done by H/W in ARMADA XP/380, and is done by S/W in ARMADA 370. the H/W BM support is not yet implemented, so all devices use the S/W management mode at this time. 1.1 Wed Jun 03 03:55:47 GMT 2015 hsuenaga branches: 1.1.2; 1.1.18; separate buffer management codes 'mvxpbm.c' from if_mvxpe.c. the buffer management(ex. fill the rx descriptors/buffers) is done by H/W in ARMADA XP/380, and is done by S/W in ARMADA 370. the H/W BM support is not yet implemented, so all devices use the S/W management mode at this time. |
| /src/common/lib/libc/arch/riscv/atomic/ | ||
| H A D | membar_ops.S | 1.4 Sat Apr 09 23:32:52 GMT 2022 riastradh Introduce membar_acquire/release. Deprecate membar_enter/exit. The names membar_enter/exit were unclear, and the documentation of membar_enter has disagreed with the implementations on sparc, powerpc, and even x86(!) for the entire time it has been in NetBSD. The terms `acquire' and `release' are ubiquitous in the literature today, and have been adopted in the C and C++ standards to mean load-before-load/store and load/store-before-store, respectively, which are exactly the orderings required by acquiring and releasing a mutex, as well as other useful applications like decrementing a reference count and then freeing the underlying object if it went to zero. Originally I proposed changing one word in the documentation for membar_enter to make it load-before-load/store instead of store-before-load/store, i.e., to make it an acquire barrier. I proposed this on the grounds that (a) all implementations guarantee load-before-load/store, (b) some implementations fail to guarantee store-before-load/store, and (c) all uses in-tree assume load-before-load/store. I verified parts (a) and (b) (except, for (a), powerpc didn't even guarantee load-before-load/store -- isync isn't necessarily enough; need lwsync in general -- but it _almost_ did, and it certainly didn't guarantee store-before-load/store). Part (c) might not be correct, however: under the mistaken assumption that atomic-r/m/w then membar-w/rw is equivalent to atomic-r/m/w then membar-r/rw, I only audited the cases of membar_enter that _aren't_ immediately after an atomic-r/m/w. All of those cases assume load-before-load/store. But my assumption was wrong -- there are cases of atomic-r/m/w then membar-w/rw that would be broken by changing to atomic-r/m/w then membar-r/rw: https://mail-index.netbsd.org/tech-kern/2022/03/29/msg028044.html Furthermore, the name membar_enter has been adopted in other places like OpenBSD where it actually does follow the documentation and guarantee store-before-load/store, even if that order is not useful. So the name membar_enter currently lives in a bad place where it means either of two things -- r/rw or w/rw. With this change, we deprecate membar_enter/exit, introduce membar_acquire/release as better names for the useful pair (r/rw and rw/w), and make sure the implementation of membar_enter guarantees both what was documented _and_ what was implemented, making it an alias for membar_sync. While here, rework all of the membar_* definitions and aliases. The new logic follows a rule to make it easier to audit: membar_X is defined as an alias for membar_Y iff membar_X is guaranteed by membar_Y. The `no stronger than' relation is (the transitive closure of): - membar_consumer (r/r) is guaranteed by membar_acquire (r/rw) - membar_producer (w/w) is guaranteed by membar_release (rw/w) - membar_acquire (r/rw) is guaranteed by membar_sync (rw/rw) - membar_release (rw/w) is guaranteed by membar_sync (rw/rw) And, for the deprecated membars: - membar_enter (whether r/rw, w/rw, or rw/rw) is guaranteed by membar_sync (rw/rw) - membar_exit (rw/w) is guaranteed by membar_release (rw/w) (membar_exit is identical to membar_release, but the name is deprecated.) Finally, while here, annotate some of the instructions with their semantics. For powerpc, leave an essay with citations on the unfortunate but -- as far as I can tell -- necessary decision to use lwsync, not isync, for membar_acquire and membar_consumer. Also add membar(3) and atomic(3) man page links. 1.4 Sat Apr 09 23:32:52 GMT 2022 riastradh Introduce membar_acquire/release. Deprecate membar_enter/exit. The names membar_enter/exit were unclear, and the documentation of membar_enter has disagreed with the implementations on sparc, powerpc, and even x86(!) for the entire time it has been in NetBSD. The terms `acquire' and `release' are ubiquitous in the literature today, and have been adopted in the C and C++ standards to mean load-before-load/store and load/store-before-store, respectively, which are exactly the orderings required by acquiring and releasing a mutex, as well as other useful applications like decrementing a reference count and then freeing the underlying object if it went to zero. Originally I proposed changing one word in the documentation for membar_enter to make it load-before-load/store instead of store-before-load/store, i.e., to make it an acquire barrier. I proposed this on the grounds that (a) all implementations guarantee load-before-load/store, (b) some implementations fail to guarantee store-before-load/store, and (c) all uses in-tree assume load-before-load/store. I verified parts (a) and (b) (except, for (a), powerpc didn't even guarantee load-before-load/store -- isync isn't necessarily enough; need lwsync in general -- but it _almost_ did, and it certainly didn't guarantee store-before-load/store). Part (c) might not be correct, however: under the mistaken assumption that atomic-r/m/w then membar-w/rw is equivalent to atomic-r/m/w then membar-r/rw, I only audited the cases of membar_enter that _aren't_ immediately after an atomic-r/m/w. All of those cases assume load-before-load/store. But my assumption was wrong -- there are cases of atomic-r/m/w then membar-w/rw that would be broken by changing to atomic-r/m/w then membar-r/rw: https://mail-index.netbsd.org/tech-kern/2022/03/29/msg028044.html Furthermore, the name membar_enter has been adopted in other places like OpenBSD where it actually does follow the documentation and guarantee store-before-load/store, even if that order is not useful. So the name membar_enter currently lives in a bad place where it means either of two things -- r/rw or w/rw. With this change, we deprecate membar_enter/exit, introduce membar_acquire/release as better names for the useful pair (r/rw and rw/w), and make sure the implementation of membar_enter guarantees both what was documented _and_ what was implemented, making it an alias for membar_sync. While here, rework all of the membar_* definitions and aliases. The new logic follows a rule to make it easier to audit: membar_X is defined as an alias for membar_Y iff membar_X is guaranteed by membar_Y. The `no stronger than' relation is (the transitive closure of): - membar_consumer (r/r) is guaranteed by membar_acquire (r/rw) - membar_producer (w/w) is guaranteed by membar_release (rw/w) - membar_acquire (r/rw) is guaranteed by membar_sync (rw/rw) - membar_release (rw/w) is guaranteed by membar_sync (rw/rw) And, for the deprecated membars: - membar_enter (whether r/rw, w/rw, or rw/rw) is guaranteed by membar_sync (rw/rw) - membar_exit (rw/w) is guaranteed by membar_release (rw/w) (membar_exit is identical to membar_release, but the name is deprecated.) Finally, while here, annotate some of the instructions with their semantics. For powerpc, leave an essay with citations on the unfortunate but -- as far as I can tell -- necessary decision to use lwsync, not isync, for membar_acquire and membar_consumer. Also add membar(3) and atomic(3) man page links. 1.4 Sat Apr 09 23:32:52 GMT 2022 riastradh Introduce membar_acquire/release. Deprecate membar_enter/exit. The names membar_enter/exit were unclear, and the documentation of membar_enter has disagreed with the implementations on sparc, powerpc, and even x86(!) for the entire time it has been in NetBSD. The terms `acquire' and `release' are ubiquitous in the literature today, and have been adopted in the C and C++ standards to mean load-before-load/store and load/store-before-store, respectively, which are exactly the orderings required by acquiring and releasing a mutex, as well as other useful applications like decrementing a reference count and then freeing the underlying object if it went to zero. Originally I proposed changing one word in the documentation for membar_enter to make it load-before-load/store instead of store-before-load/store, i.e., to make it an acquire barrier. I proposed this on the grounds that (a) all implementations guarantee load-before-load/store, (b) some implementations fail to guarantee store-before-load/store, and (c) all uses in-tree assume load-before-load/store. I verified parts (a) and (b) (except, for (a), powerpc didn't even guarantee load-before-load/store -- isync isn't necessarily enough; need lwsync in general -- but it _almost_ did, and it certainly didn't guarantee store-before-load/store). Part (c) might not be correct, however: under the mistaken assumption that atomic-r/m/w then membar-w/rw is equivalent to atomic-r/m/w then membar-r/rw, I only audited the cases of membar_enter that _aren't_ immediately after an atomic-r/m/w. All of those cases assume load-before-load/store. But my assumption was wrong -- there are cases of atomic-r/m/w then membar-w/rw that would be broken by changing to atomic-r/m/w then membar-r/rw: https://mail-index.netbsd.org/tech-kern/2022/03/29/msg028044.html Furthermore, the name membar_enter has been adopted in other places like OpenBSD where it actually does follow the documentation and guarantee store-before-load/store, even if that order is not useful. So the name membar_enter currently lives in a bad place where it means either of two things -- r/rw or w/rw. With this change, we deprecate membar_enter/exit, introduce membar_acquire/release as better names for the useful pair (r/rw and rw/w), and make sure the implementation of membar_enter guarantees both what was documented _and_ what was implemented, making it an alias for membar_sync. While here, rework all of the membar_* definitions and aliases. The new logic follows a rule to make it easier to audit: membar_X is defined as an alias for membar_Y iff membar_X is guaranteed by membar_Y. The `no stronger than' relation is (the transitive closure of): - membar_consumer (r/r) is guaranteed by membar_acquire (r/rw) - membar_producer (w/w) is guaranteed by membar_release (rw/w) - membar_acquire (r/rw) is guaranteed by membar_sync (rw/rw) - membar_release (rw/w) is guaranteed by membar_sync (rw/rw) And, for the deprecated membars: - membar_enter (whether r/rw, w/rw, or rw/rw) is guaranteed by membar_sync (rw/rw) - membar_exit (rw/w) is guaranteed by membar_release (rw/w) (membar_exit is identical to membar_release, but the name is deprecated.) Finally, while here, annotate some of the instructions with their semantics. For powerpc, leave an essay with citations on the unfortunate but -- as far as I can tell -- necessary decision to use lwsync, not isync, for membar_acquire and membar_consumer. Also add membar(3) and atomic(3) man page links. 1.4 Sat Apr 09 23:32:52 GMT 2022 riastradh Introduce membar_acquire/release. Deprecate membar_enter/exit. The names membar_enter/exit were unclear, and the documentation of membar_enter has disagreed with the implementations on sparc, powerpc, and even x86(!) for the entire time it has been in NetBSD. The terms `acquire' and `release' are ubiquitous in the literature today, and have been adopted in the C and C++ standards to mean load-before-load/store and load/store-before-store, respectively, which are exactly the orderings required by acquiring and releasing a mutex, as well as other useful applications like decrementing a reference count and then freeing the underlying object if it went to zero. Originally I proposed changing one word in the documentation for membar_enter to make it load-before-load/store instead of store-before-load/store, i.e., to make it an acquire barrier. I proposed this on the grounds that (a) all implementations guarantee load-before-load/store, (b) some implementations fail to guarantee store-before-load/store, and (c) all uses in-tree assume load-before-load/store. I verified parts (a) and (b) (except, for (a), powerpc didn't even guarantee load-before-load/store -- isync isn't necessarily enough; need lwsync in general -- but it _almost_ did, and it certainly didn't guarantee store-before-load/store). Part (c) might not be correct, however: under the mistaken assumption that atomic-r/m/w then membar-w/rw is equivalent to atomic-r/m/w then membar-r/rw, I only audited the cases of membar_enter that _aren't_ immediately after an atomic-r/m/w. All of those cases assume load-before-load/store. But my assumption was wrong -- there are cases of atomic-r/m/w then membar-w/rw that would be broken by changing to atomic-r/m/w then membar-r/rw: https://mail-index.netbsd.org/tech-kern/2022/03/29/msg028044.html Furthermore, the name membar_enter has been adopted in other places like OpenBSD where it actually does follow the documentation and guarantee store-before-load/store, even if that order is not useful. So the name membar_enter currently lives in a bad place where it means either of two things -- r/rw or w/rw. With this change, we deprecate membar_enter/exit, introduce membar_acquire/release as better names for the useful pair (r/rw and rw/w), and make sure the implementation of membar_enter guarantees both what was documented _and_ what was implemented, making it an alias for membar_sync. While here, rework all of the membar_* definitions and aliases. The new logic follows a rule to make it easier to audit: membar_X is defined as an alias for membar_Y iff membar_X is guaranteed by membar_Y. The `no stronger than' relation is (the transitive closure of): - membar_consumer (r/r) is guaranteed by membar_acquire (r/rw) - membar_producer (w/w) is guaranteed by membar_release (rw/w) - membar_acquire (r/rw) is guaranteed by membar_sync (rw/rw) - membar_release (rw/w) is guaranteed by membar_sync (rw/rw) And, for the deprecated membars: - membar_enter (whether r/rw, w/rw, or rw/rw) is guaranteed by membar_sync (rw/rw) - membar_exit (rw/w) is guaranteed by membar_release (rw/w) (membar_exit is identical to membar_release, but the name is deprecated.) Finally, while here, annotate some of the instructions with their semantics. For powerpc, leave an essay with citations on the unfortunate but -- as far as I can tell -- necessary decision to use lwsync, not isync, for membar_acquire and membar_consumer. Also add membar(3) and atomic(3) man page links. 1.4 Sat Apr 09 23:32:52 GMT 2022 riastradh Introduce membar_acquire/release. Deprecate membar_enter/exit. The names membar_enter/exit were unclear, and the documentation of membar_enter has disagreed with the implementations on sparc, powerpc, and even x86(!) for the entire time it has been in NetBSD. The terms `acquire' and `release' are ubiquitous in the literature today, and have been adopted in the C and C++ standards to mean load-before-load/store and load/store-before-store, respectively, which are exactly the orderings required by acquiring and releasing a mutex, as well as other useful applications like decrementing a reference count and then freeing the underlying object if it went to zero. Originally I proposed changing one word in the documentation for membar_enter to make it load-before-load/store instead of store-before-load/store, i.e., to make it an acquire barrier. I proposed this on the grounds that (a) all implementations guarantee load-before-load/store, (b) some implementations fail to guarantee store-before-load/store, and (c) all uses in-tree assume load-before-load/store. I verified parts (a) and (b) (except, for (a), powerpc didn't even guarantee load-before-load/store -- isync isn't necessarily enough; need lwsync in general -- but it _almost_ did, and it certainly didn't guarantee store-before-load/store). Part (c) might not be correct, however: under the mistaken assumption that atomic-r/m/w then membar-w/rw is equivalent to atomic-r/m/w then membar-r/rw, I only audited the cases of membar_enter that _aren't_ immediately after an atomic-r/m/w. All of those cases assume load-before-load/store. But my assumption was wrong -- there are cases of atomic-r/m/w then membar-w/rw that would be broken by changing to atomic-r/m/w then membar-r/rw: https://mail-index.netbsd.org/tech-kern/2022/03/29/msg028044.html Furthermore, the name membar_enter has been adopted in other places like OpenBSD where it actually does follow the documentation and guarantee store-before-load/store, even if that order is not useful. So the name membar_enter currently lives in a bad place where it means either of two things -- r/rw or w/rw. With this change, we deprecate membar_enter/exit, introduce membar_acquire/release as better names for the useful pair (r/rw and rw/w), and make sure the implementation of membar_enter guarantees both what was documented _and_ what was implemented, making it an alias for membar_sync. While here, rework all of the membar_* definitions and aliases. The new logic follows a rule to make it easier to audit: membar_X is defined as an alias for membar_Y iff membar_X is guaranteed by membar_Y. The `no stronger than' relation is (the transitive closure of): - membar_consumer (r/r) is guaranteed by membar_acquire (r/rw) - membar_producer (w/w) is guaranteed by membar_release (rw/w) - membar_acquire (r/rw) is guaranteed by membar_sync (rw/rw) - membar_release (rw/w) is guaranteed by membar_sync (rw/rw) And, for the deprecated membars: - membar_enter (whether r/rw, w/rw, or rw/rw) is guaranteed by membar_sync (rw/rw) - membar_exit (rw/w) is guaranteed by membar_release (rw/w) (membar_exit is identical to membar_release, but the name is deprecated.) Finally, while here, annotate some of the instructions with their semantics. For powerpc, leave an essay with citations on the unfortunate but -- as far as I can tell -- necessary decision to use lwsync, not isync, for membar_acquire and membar_consumer. Also add membar(3) and atomic(3) man page links. 1.4 Sat Apr 09 23:32:52 GMT 2022 riastradh Introduce membar_acquire/release. Deprecate membar_enter/exit. The names membar_enter/exit were unclear, and the documentation of membar_enter has disagreed with the implementations on sparc, powerpc, and even x86(!) for the entire time it has been in NetBSD. The terms `acquire' and `release' are ubiquitous in the literature today, and have been adopted in the C and C++ standards to mean load-before-load/store and load/store-before-store, respectively, which are exactly the orderings required by acquiring and releasing a mutex, as well as other useful applications like decrementing a reference count and then freeing the underlying object if it went to zero. Originally I proposed changing one word in the documentation for membar_enter to make it load-before-load/store instead of store-before-load/store, i.e., to make it an acquire barrier. I proposed this on the grounds that (a) all implementations guarantee load-before-load/store, (b) some implementations fail to guarantee store-before-load/store, and (c) all uses in-tree assume load-before-load/store. I verified parts (a) and (b) (except, for (a), powerpc didn't even guarantee load-before-load/store -- isync isn't necessarily enough; need lwsync in general -- but it _almost_ did, and it certainly didn't guarantee store-before-load/store). Part (c) might not be correct, however: under the mistaken assumption that atomic-r/m/w then membar-w/rw is equivalent to atomic-r/m/w then membar-r/rw, I only audited the cases of membar_enter that _aren't_ immediately after an atomic-r/m/w. All of those cases assume load-before-load/store. But my assumption was wrong -- there are cases of atomic-r/m/w then membar-w/rw that would be broken by changing to atomic-r/m/w then membar-r/rw: https://mail-index.netbsd.org/tech-kern/2022/03/29/msg028044.html Furthermore, the name membar_enter has been adopted in other places like OpenBSD where it actually does follow the documentation and guarantee store-before-load/store, even if that order is not useful. So the name membar_enter currently lives in a bad place where it means either of two things -- r/rw or w/rw. With this change, we deprecate membar_enter/exit, introduce membar_acquire/release as better names for the useful pair (r/rw and rw/w), and make sure the implementation of membar_enter guarantees both what was documented _and_ what was implemented, making it an alias for membar_sync. While here, rework all of the membar_* definitions and aliases. The new logic follows a rule to make it easier to audit: membar_X is defined as an alias for membar_Y iff membar_X is guaranteed by membar_Y. The `no stronger than' relation is (the transitive closure of): - membar_consumer (r/r) is guaranteed by membar_acquire (r/rw) - membar_producer (w/w) is guaranteed by membar_release (rw/w) - membar_acquire (r/rw) is guaranteed by membar_sync (rw/rw) - membar_release (rw/w) is guaranteed by membar_sync (rw/rw) And, for the deprecated membars: - membar_enter (whether r/rw, w/rw, or rw/rw) is guaranteed by membar_sync (rw/rw) - membar_exit (rw/w) is guaranteed by membar_release (rw/w) (membar_exit is identical to membar_release, but the name is deprecated.) Finally, while here, annotate some of the instructions with their semantics. For powerpc, leave an essay with citations on the unfortunate but -- as far as I can tell -- necessary decision to use lwsync, not isync, for membar_acquire and membar_consumer. Also add membar(3) and atomic(3) man page links. 1.4 Sat Apr 09 23:32:52 GMT 2022 riastradh Introduce membar_acquire/release. Deprecate membar_enter/exit. The names membar_enter/exit were unclear, and the documentation of membar_enter has disagreed with the implementations on sparc, powerpc, and even x86(!) for the entire time it has been in NetBSD. The terms `acquire' and `release' are ubiquitous in the literature today, and have been adopted in the C and C++ standards to mean load-before-load/store and load/store-before-store, respectively, which are exactly the orderings required by acquiring and releasing a mutex, as well as other useful applications like decrementing a reference count and then freeing the underlying object if it went to zero. Originally I proposed changing one word in the documentation for membar_enter to make it load-before-load/store instead of store-before-load/store, i.e., to make it an acquire barrier. I proposed this on the grounds that (a) all implementations guarantee load-before-load/store, (b) some implementations fail to guarantee store-before-load/store, and (c) all uses in-tree assume load-before-load/store. I verified parts (a) and (b) (except, for (a), powerpc didn't even guarantee load-before-load/store -- isync isn't necessarily enough; need lwsync in general -- but it _almost_ did, and it certainly didn't guarantee store-before-load/store). Part (c) might not be correct, however: under the mistaken assumption that atomic-r/m/w then membar-w/rw is equivalent to atomic-r/m/w then membar-r/rw, I only audited the cases of membar_enter that _aren't_ immediately after an atomic-r/m/w. All of those cases assume load-before-load/store. But my assumption was wrong -- there are cases of atomic-r/m/w then membar-w/rw that would be broken by changing to atomic-r/m/w then membar-r/rw: https://mail-index.netbsd.org/tech-kern/2022/03/29/msg028044.html Furthermore, the name membar_enter has been adopted in other places like OpenBSD where it actually does follow the documentation and guarantee store-before-load/store, even if that order is not useful. So the name membar_enter currently lives in a bad place where it means either of two things -- r/rw or w/rw. With this change, we deprecate membar_enter/exit, introduce membar_acquire/release as better names for the useful pair (r/rw and rw/w), and make sure the implementation of membar_enter guarantees both what was documented _and_ what was implemented, making it an alias for membar_sync. While here, rework all of the membar_* definitions and aliases. The new logic follows a rule to make it easier to audit: membar_X is defined as an alias for membar_Y iff membar_X is guaranteed by membar_Y. The `no stronger than' relation is (the transitive closure of): - membar_consumer (r/r) is guaranteed by membar_acquire (r/rw) - membar_producer (w/w) is guaranteed by membar_release (rw/w) - membar_acquire (r/rw) is guaranteed by membar_sync (rw/rw) - membar_release (rw/w) is guaranteed by membar_sync (rw/rw) And, for the deprecated membars: - membar_enter (whether r/rw, w/rw, or rw/rw) is guaranteed by membar_sync (rw/rw) - membar_exit (rw/w) is guaranteed by membar_release (rw/w) (membar_exit is identical to membar_release, but the name is deprecated.) Finally, while here, annotate some of the instructions with their semantics. For powerpc, leave an essay with citations on the unfortunate but -- as far as I can tell -- necessary decision to use lwsync, not isync, for membar_acquire and membar_consumer. Also add membar(3) and atomic(3) man page links. 1.4 Sat Apr 09 23:32:52 GMT 2022 riastradh Introduce membar_acquire/release. Deprecate membar_enter/exit. The names membar_enter/exit were unclear, and the documentation of membar_enter has disagreed with the implementations on sparc, powerpc, and even x86(!) for the entire time it has been in NetBSD. The terms `acquire' and `release' are ubiquitous in the literature today, and have been adopted in the C and C++ standards to mean load-before-load/store and load/store-before-store, respectively, which are exactly the orderings required by acquiring and releasing a mutex, as well as other useful applications like decrementing a reference count and then freeing the underlying object if it went to zero. Originally I proposed changing one word in the documentation for membar_enter to make it load-before-load/store instead of store-before-load/store, i.e., to make it an acquire barrier. I proposed this on the grounds that (a) all implementations guarantee load-before-load/store, (b) some implementations fail to guarantee store-before-load/store, and (c) all uses in-tree assume load-before-load/store. I verified parts (a) and (b) (except, for (a), powerpc didn't even guarantee load-before-load/store -- isync isn't necessarily enough; need lwsync in general -- but it _almost_ did, and it certainly didn't guarantee store-before-load/store). Part (c) might not be correct, however: under the mistaken assumption that atomic-r/m/w then membar-w/rw is equivalent to atomic-r/m/w then membar-r/rw, I only audited the cases of membar_enter that _aren't_ immediately after an atomic-r/m/w. All of those cases assume load-before-load/store. But my assumption was wrong -- there are cases of atomic-r/m/w then membar-w/rw that would be broken by changing to atomic-r/m/w then membar-r/rw: https://mail-index.netbsd.org/tech-kern/2022/03/29/msg028044.html Furthermore, the name membar_enter has been adopted in other places like OpenBSD where it actually does follow the documentation and guarantee store-before-load/store, even if that order is not useful. So the name membar_enter currently lives in a bad place where it means either of two things -- r/rw or w/rw. With this change, we deprecate membar_enter/exit, introduce membar_acquire/release as better names for the useful pair (r/rw and rw/w), and make sure the implementation of membar_enter guarantees both what was documented _and_ what was implemented, making it an alias for membar_sync. While here, rework all of the membar_* definitions and aliases. The new logic follows a rule to make it easier to audit: membar_X is defined as an alias for membar_Y iff membar_X is guaranteed by membar_Y. The `no stronger than' relation is (the transitive closure of): - membar_consumer (r/r) is guaranteed by membar_acquire (r/rw) - membar_producer (w/w) is guaranteed by membar_release (rw/w) - membar_acquire (r/rw) is guaranteed by membar_sync (rw/rw) - membar_release (rw/w) is guaranteed by membar_sync (rw/rw) And, for the deprecated membars: - membar_enter (whether r/rw, w/rw, or rw/rw) is guaranteed by membar_sync (rw/rw) - membar_exit (rw/w) is guaranteed by membar_release (rw/w) (membar_exit is identical to membar_release, but the name is deprecated.) Finally, while here, annotate some of the instructions with their semantics. For powerpc, leave an essay with citations on the unfortunate but -- as far as I can tell -- necessary decision to use lwsync, not isync, for membar_acquire and membar_consumer. Also add membar(3) and atomic(3) man page links. 1.4 Sat Apr 09 23:32:52 GMT 2022 riastradh Introduce membar_acquire/release. Deprecate membar_enter/exit. The names membar_enter/exit were unclear, and the documentation of membar_enter has disagreed with the implementations on sparc, powerpc, and even x86(!) for the entire time it has been in NetBSD. The terms `acquire' and `release' are ubiquitous in the literature today, and have been adopted in the C and C++ standards to mean load-before-load/store and load/store-before-store, respectively, which are exactly the orderings required by acquiring and releasing a mutex, as well as other useful applications like decrementing a reference count and then freeing the underlying object if it went to zero. Originally I proposed changing one word in the documentation for membar_enter to make it load-before-load/store instead of store-before-load/store, i.e., to make it an acquire barrier. I proposed this on the grounds that (a) all implementations guarantee load-before-load/store, (b) some implementations fail to guarantee store-before-load/store, and (c) all uses in-tree assume load-before-load/store. I verified parts (a) and (b) (except, for (a), powerpc didn't even guarantee load-before-load/store -- isync isn't necessarily enough; need lwsync in general -- but it _almost_ did, and it certainly didn't guarantee store-before-load/store). Part (c) might not be correct, however: under the mistaken assumption that atomic-r/m/w then membar-w/rw is equivalent to atomic-r/m/w then membar-r/rw, I only audited the cases of membar_enter that _aren't_ immediately after an atomic-r/m/w. All of those cases assume load-before-load/store. But my assumption was wrong -- there are cases of atomic-r/m/w then membar-w/rw that would be broken by changing to atomic-r/m/w then membar-r/rw: https://mail-index.netbsd.org/tech-kern/2022/03/29/msg028044.html Furthermore, the name membar_enter has been adopted in other places like OpenBSD where it actually does follow the documentation and guarantee store-before-load/store, even if that order is not useful. So the name membar_enter currently lives in a bad place where it means either of two things -- r/rw or w/rw. With this change, we deprecate membar_enter/exit, introduce membar_acquire/release as better names for the useful pair (r/rw and rw/w), and make sure the implementation of membar_enter guarantees both what was documented _and_ what was implemented, making it an alias for membar_sync. While here, rework all of the membar_* definitions and aliases. The new logic follows a rule to make it easier to audit: membar_X is defined as an alias for membar_Y iff membar_X is guaranteed by membar_Y. The `no stronger than' relation is (the transitive closure of): - membar_consumer (r/r) is guaranteed by membar_acquire (r/rw) - membar_producer (w/w) is guaranteed by membar_release (rw/w) - membar_acquire (r/rw) is guaranteed by membar_sync (rw/rw) - membar_release (rw/w) is guaranteed by membar_sync (rw/rw) And, for the deprecated membars: - membar_enter (whether r/rw, w/rw, or rw/rw) is guaranteed by membar_sync (rw/rw) - membar_exit (rw/w) is guaranteed by membar_release (rw/w) (membar_exit is identical to membar_release, but the name is deprecated.) Finally, while here, annotate some of the instructions with their semantics. For powerpc, leave an essay with citations on the unfortunate but -- as far as I can tell -- necessary decision to use lwsync, not isync, for membar_acquire and membar_consumer. Also add membar(3) and atomic(3) man page links. 1.4 Sat Apr 09 23:32:52 GMT 2022 riastradh Introduce membar_acquire/release. Deprecate membar_enter/exit. The names membar_enter/exit were unclear, and the documentation of membar_enter has disagreed with the implementations on sparc, powerpc, and even x86(!) for the entire time it has been in NetBSD. The terms `acquire' and `release' are ubiquitous in the literature today, and have been adopted in the C and C++ standards to mean load-before-load/store and load/store-before-store, respectively, which are exactly the orderings required by acquiring and releasing a mutex, as well as other useful applications like decrementing a reference count and then freeing the underlying object if it went to zero. Originally I proposed changing one word in the documentation for membar_enter to make it load-before-load/store instead of store-before-load/store, i.e., to make it an acquire barrier. I proposed this on the grounds that (a) all implementations guarantee load-before-load/store, (b) some implementations fail to guarantee store-before-load/store, and (c) all uses in-tree assume load-before-load/store. I verified parts (a) and (b) (except, for (a), powerpc didn't even guarantee load-before-load/store -- isync isn't necessarily enough; need lwsync in general -- but it _almost_ did, and it certainly didn't guarantee store-before-load/store). Part (c) might not be correct, however: under the mistaken assumption that atomic-r/m/w then membar-w/rw is equivalent to atomic-r/m/w then membar-r/rw, I only audited the cases of membar_enter that _aren't_ immediately after an atomic-r/m/w. All of those cases assume load-before-load/store. But my assumption was wrong -- there are cases of atomic-r/m/w then membar-w/rw that would be broken by changing to atomic-r/m/w then membar-r/rw: https://mail-index.netbsd.org/tech-kern/2022/03/29/msg028044.html Furthermore, the name membar_enter has been adopted in other places like OpenBSD where it actually does follow the documentation and guarantee store-before-load/store, even if that order is not useful. So the name membar_enter currently lives in a bad place where it means either of two things -- r/rw or w/rw. With this change, we deprecate membar_enter/exit, introduce membar_acquire/release as better names for the useful pair (r/rw and rw/w), and make sure the implementation of membar_enter guarantees both what was documented _and_ what was implemented, making it an alias for membar_sync. While here, rework all of the membar_* definitions and aliases. The new logic follows a rule to make it easier to audit: membar_X is defined as an alias for membar_Y iff membar_X is guaranteed by membar_Y. The `no stronger than' relation is (the transitive closure of): - membar_consumer (r/r) is guaranteed by membar_acquire (r/rw) - membar_producer (w/w) is guaranteed by membar_release (rw/w) - membar_acquire (r/rw) is guaranteed by membar_sync (rw/rw) - membar_release (rw/w) is guaranteed by membar_sync (rw/rw) And, for the deprecated membars: - membar_enter (whether r/rw, w/rw, or rw/rw) is guaranteed by membar_sync (rw/rw) - membar_exit (rw/w) is guaranteed by membar_release (rw/w) (membar_exit is identical to membar_release, but the name is deprecated.) Finally, while here, annotate some of the instructions with their semantics. For powerpc, leave an essay with citations on the unfortunate but -- as far as I can tell -- necessary decision to use lwsync, not isync, for membar_acquire and membar_consumer. Also add membar(3) and atomic(3) man page links. 1.4 Sat Apr 09 23:32:52 GMT 2022 riastradh Introduce membar_acquire/release. Deprecate membar_enter/exit. The names membar_enter/exit were unclear, and the documentation of membar_enter has disagreed with the implementations on sparc, powerpc, and even x86(!) for the entire time it has been in NetBSD. The terms `acquire' and `release' are ubiquitous in the literature today, and have been adopted in the C and C++ standards to mean load-before-load/store and load/store-before-store, respectively, which are exactly the orderings required by acquiring and releasing a mutex, as well as other useful applications like decrementing a reference count and then freeing the underlying object if it went to zero. Originally I proposed changing one word in the documentation for membar_enter to make it load-before-load/store instead of store-before-load/store, i.e., to make it an acquire barrier. I proposed this on the grounds that (a) all implementations guarantee load-before-load/store, (b) some implementations fail to guarantee store-before-load/store, and (c) all uses in-tree assume load-before-load/store. I verified parts (a) and (b) (except, for (a), powerpc didn't even guarantee load-before-load/store -- isync isn't necessarily enough; need lwsync in general -- but it _almost_ did, and it certainly didn't guarantee store-before-load/store). Part (c) might not be correct, however: under the mistaken assumption that atomic-r/m/w then membar-w/rw is equivalent to atomic-r/m/w then membar-r/rw, I only audited the cases of membar_enter that _aren't_ immediately after an atomic-r/m/w. All of those cases assume load-before-load/store. But my assumption was wrong -- there are cases of atomic-r/m/w then membar-w/rw that would be broken by changing to atomic-r/m/w then membar-r/rw: https://mail-index.netbsd.org/tech-kern/2022/03/29/msg028044.html Furthermore, the name membar_enter has been adopted in other places like OpenBSD where it actually does follow the documentation and guarantee store-before-load/store, even if that order is not useful. So the name membar_enter currently lives in a bad place where it means either of two things -- r/rw or w/rw. With this change, we deprecate membar_enter/exit, introduce membar_acquire/release as better names for the useful pair (r/rw and rw/w), and make sure the implementation of membar_enter guarantees both what was documented _and_ what was implemented, making it an alias for membar_sync. While here, rework all of the membar_* definitions and aliases. The new logic follows a rule to make it easier to audit: membar_X is defined as an alias for membar_Y iff membar_X is guaranteed by membar_Y. The `no stronger than' relation is (the transitive closure of): - membar_consumer (r/r) is guaranteed by membar_acquire (r/rw) - membar_producer (w/w) is guaranteed by membar_release (rw/w) - membar_acquire (r/rw) is guaranteed by membar_sync (rw/rw) - membar_release (rw/w) is guaranteed by membar_sync (rw/rw) And, for the deprecated membars: - membar_enter (whether r/rw, w/rw, or rw/rw) is guaranteed by membar_sync (rw/rw) - membar_exit (rw/w) is guaranteed by membar_release (rw/w) (membar_exit is identical to membar_release, but the name is deprecated.) Finally, while here, annotate some of the instructions with their semantics. For powerpc, leave an essay with citations on the unfortunate but -- as far as I can tell -- necessary decision to use lwsync, not isync, for membar_acquire and membar_consumer. Also add membar(3) and atomic(3) man page links. 1.4 Sat Apr 09 23:32:52 GMT 2022 riastradh Introduce membar_acquire/release. Deprecate membar_enter/exit. The names membar_enter/exit were unclear, and the documentation of membar_enter has disagreed with the implementations on sparc, powerpc, and even x86(!) for the entire time it has been in NetBSD. The terms `acquire' and `release' are ubiquitous in the literature today, and have been adopted in the C and C++ standards to mean load-before-load/store and load/store-before-store, respectively, which are exactly the orderings required by acquiring and releasing a mutex, as well as other useful applications like decrementing a reference count and then freeing the underlying object if it went to zero. Originally I proposed changing one word in the documentation for membar_enter to make it load-before-load/store instead of store-before-load/store, i.e., to make it an acquire barrier. I proposed this on the grounds that (a) all implementations guarantee load-before-load/store, (b) some implementations fail to guarantee store-before-load/store, and (c) all uses in-tree assume load-before-load/store. I verified parts (a) and (b) (except, for (a), powerpc didn't even guarantee load-before-load/store -- isync isn't necessarily enough; need lwsync in general -- but it _almost_ did, and it certainly didn't guarantee store-before-load/store). Part (c) might not be correct, however: under the mistaken assumption that atomic-r/m/w then membar-w/rw is equivalent to atomic-r/m/w then membar-r/rw, I only audited the cases of membar_enter that _aren't_ immediately after an atomic-r/m/w. All of those cases assume load-before-load/store. But my assumption was wrong -- there are cases of atomic-r/m/w then membar-w/rw that would be broken by changing to atomic-r/m/w then membar-r/rw: https://mail-index.netbsd.org/tech-kern/2022/03/29/msg028044.html Furthermore, the name membar_enter has been adopted in other places like OpenBSD where it actually does follow the documentation and guarantee store-before-load/store, even if that order is not useful. So the name membar_enter currently lives in a bad place where it means either of two things -- r/rw or w/rw. With this change, we deprecate membar_enter/exit, introduce membar_acquire/release as better names for the useful pair (r/rw and rw/w), and make sure the implementation of membar_enter guarantees both what was documented _and_ what was implemented, making it an alias for membar_sync. While here, rework all of the membar_* definitions and aliases. The new logic follows a rule to make it easier to audit: membar_X is defined as an alias for membar_Y iff membar_X is guaranteed by membar_Y. The `no stronger than' relation is (the transitive closure of): - membar_consumer (r/r) is guaranteed by membar_acquire (r/rw) - membar_producer (w/w) is guaranteed by membar_release (rw/w) - membar_acquire (r/rw) is guaranteed by membar_sync (rw/rw) - membar_release (rw/w) is guaranteed by membar_sync (rw/rw) And, for the deprecated membars: - membar_enter (whether r/rw, w/rw, or rw/rw) is guaranteed by membar_sync (rw/rw) - membar_exit (rw/w) is guaranteed by membar_release (rw/w) (membar_exit is identical to membar_release, but the name is deprecated.) Finally, while here, annotate some of the instructions with their semantics. For powerpc, leave an essay with citations on the unfortunate but -- as far as I can tell -- necessary decision to use lwsync, not isync, for membar_acquire and membar_consumer. Also add membar(3) and atomic(3) man page links. 1.4 Sat Apr 09 23:32:52 GMT 2022 riastradh Introduce membar_acquire/release. Deprecate membar_enter/exit. The names membar_enter/exit were unclear, and the documentation of membar_enter has disagreed with the implementations on sparc, powerpc, and even x86(!) for the entire time it has been in NetBSD. The terms `acquire' and `release' are ubiquitous in the literature today, and have been adopted in the C and C++ standards to mean load-before-load/store and load/store-before-store, respectively, which are exactly the orderings required by acquiring and releasing a mutex, as well as other useful applications like decrementing a reference count and then freeing the underlying object if it went to zero. Originally I proposed changing one word in the documentation for membar_enter to make it load-before-load/store instead of store-before-load/store, i.e., to make it an acquire barrier. I proposed this on the grounds that (a) all implementations guarantee load-before-load/store, (b) some implementations fail to guarantee store-before-load/store, and (c) all uses in-tree assume load-before-load/store. I verified parts (a) and (b) (except, for (a), powerpc didn't even guarantee load-before-load/store -- isync isn't necessarily enough; need lwsync in general -- but it _almost_ did, and it certainly didn't guarantee store-before-load/store). Part (c) might not be correct, however: under the mistaken assumption that atomic-r/m/w then membar-w/rw is equivalent to atomic-r/m/w then membar-r/rw, I only audited the cases of membar_enter that _aren't_ immediately after an atomic-r/m/w. All of those cases assume load-before-load/store. But my assumption was wrong -- there are cases of atomic-r/m/w then membar-w/rw that would be broken by changing to atomic-r/m/w then membar-r/rw: https://mail-index.netbsd.org/tech-kern/2022/03/29/msg028044.html Furthermore, the name membar_enter has been adopted in other places like OpenBSD where it actually does follow the documentation and guarantee store-before-load/store, even if that order is not useful. So the name membar_enter currently lives in a bad place where it means either of two things -- r/rw or w/rw. With this change, we deprecate membar_enter/exit, introduce membar_acquire/release as better names for the useful pair (r/rw and rw/w), and make sure the implementation of membar_enter guarantees both what was documented _and_ what was implemented, making it an alias for membar_sync. While here, rework all of the membar_* definitions and aliases. The new logic follows a rule to make it easier to audit: membar_X is defined as an alias for membar_Y iff membar_X is guaranteed by membar_Y. The `no stronger than' relation is (the transitive closure of): - membar_consumer (r/r) is guaranteed by membar_acquire (r/rw) - membar_producer (w/w) is guaranteed by membar_release (rw/w) - membar_acquire (r/rw) is guaranteed by membar_sync (rw/rw) - membar_release (rw/w) is guaranteed by membar_sync (rw/rw) And, for the deprecated membars: - membar_enter (whether r/rw, w/rw, or rw/rw) is guaranteed by membar_sync (rw/rw) - membar_exit (rw/w) is guaranteed by membar_release (rw/w) (membar_exit is identical to membar_release, but the name is deprecated.) Finally, while here, annotate some of the instructions with their semantics. For powerpc, leave an essay with citations on the unfortunate but -- as far as I can tell -- necessary decision to use lwsync, not isync, for membar_acquire and membar_consumer. Also add membar(3) and atomic(3) man page links. 1.4 Sat Apr 09 23:32:52 GMT 2022 riastradh Introduce membar_acquire/release. Deprecate membar_enter/exit. The names membar_enter/exit were unclear, and the documentation of membar_enter has disagreed with the implementations on sparc, powerpc, and even x86(!) for the entire time it has been in NetBSD. The terms `acquire' and `release' are ubiquitous in the literature today, and have been adopted in the C and C++ standards to mean load-before-load/store and load/store-before-store, respectively, which are exactly the orderings required by acquiring and releasing a mutex, as well as other useful applications like decrementing a reference count and then freeing the underlying object if it went to zero. Originally I proposed changing one word in the documentation for membar_enter to make it load-before-load/store instead of store-before-load/store, i.e., to make it an acquire barrier. I proposed this on the grounds that (a) all implementations guarantee load-before-load/store, (b) some implementations fail to guarantee store-before-load/store, and (c) all uses in-tree assume load-before-load/store. I verified parts (a) and (b) (except, for (a), powerpc didn't even guarantee load-before-load/store -- isync isn't necessarily enough; need lwsync in general -- but it _almost_ did, and it certainly didn't guarantee store-before-load/store). Part (c) might not be correct, however: under the mistaken assumption that atomic-r/m/w then membar-w/rw is equivalent to atomic-r/m/w then membar-r/rw, I only audited the cases of membar_enter that _aren't_ immediately after an atomic-r/m/w. All of those cases assume load-before-load/store. But my assumption was wrong -- there are cases of atomic-r/m/w then membar-w/rw that would be broken by changing to atomic-r/m/w then membar-r/rw: https://mail-index.netbsd.org/tech-kern/2022/03/29/msg028044.html Furthermore, the name membar_enter has been adopted in other places like OpenBSD where it actually does follow the documentation and guarantee store-before-load/store, even if that order is not useful. So the name membar_enter currently lives in a bad place where it means either of two things -- r/rw or w/rw. With this change, we deprecate membar_enter/exit, introduce membar_acquire/release as better names for the useful pair (r/rw and rw/w), and make sure the implementation of membar_enter guarantees both what was documented _and_ what was implemented, making it an alias for membar_sync. While here, rework all of the membar_* definitions and aliases. The new logic follows a rule to make it easier to audit: membar_X is defined as an alias for membar_Y iff membar_X is guaranteed by membar_Y. The `no stronger than' relation is (the transitive closure of): - membar_consumer (r/r) is guaranteed by membar_acquire (r/rw) - membar_producer (w/w) is guaranteed by membar_release (rw/w) - membar_acquire (r/rw) is guaranteed by membar_sync (rw/rw) - membar_release (rw/w) is guaranteed by membar_sync (rw/rw) And, for the deprecated membars: - membar_enter (whether r/rw, w/rw, or rw/rw) is guaranteed by membar_sync (rw/rw) - membar_exit (rw/w) is guaranteed by membar_release (rw/w) (membar_exit is identical to membar_release, but the name is deprecated.) Finally, while here, annotate some of the instructions with their semantics. For powerpc, leave an essay with citations on the unfortunate but -- as far as I can tell -- necessary decision to use lwsync, not isync, for membar_acquire and membar_consumer. Also add membar(3) and atomic(3) man page links. 1.4 Sat Apr 09 23:32:52 GMT 2022 riastradh Introduce membar_acquire/release. Deprecate membar_enter/exit. The names membar_enter/exit were unclear, and the documentation of membar_enter has disagreed with the implementations on sparc, powerpc, and even x86(!) for the entire time it has been in NetBSD. The terms `acquire' and `release' are ubiquitous in the literature today, and have been adopted in the C and C++ standards to mean load-before-load/store and load/store-before-store, respectively, which are exactly the orderings required by acquiring and releasing a mutex, as well as other useful applications like decrementing a reference count and then freeing the underlying object if it went to zero. Originally I proposed changing one word in the documentation for membar_enter to make it load-before-load/store instead of store-before-load/store, i.e., to make it an acquire barrier. I proposed this on the grounds that (a) all implementations guarantee load-before-load/store, (b) some implementations fail to guarantee store-before-load/store, and (c) all uses in-tree assume load-before-load/store. I verified parts (a) and (b) (except, for (a), powerpc didn't even guarantee load-before-load/store -- isync isn't necessarily enough; need lwsync in general -- but it _almost_ did, and it certainly didn't guarantee store-before-load/store). Part (c) might not be correct, however: under the mistaken assumption that atomic-r/m/w then membar-w/rw is equivalent to atomic-r/m/w then membar-r/rw, I only audited the cases of membar_enter that _aren't_ immediately after an atomic-r/m/w. All of those cases assume load-before-load/store. But my assumption was wrong -- there are cases of atomic-r/m/w then membar-w/rw that would be broken by changing to atomic-r/m/w then membar-r/rw: https://mail-index.netbsd.org/tech-kern/2022/03/29/msg028044.html Furthermore, the name membar_enter has been adopted in other places like OpenBSD where it actually does follow the documentation and guarantee store-before-load/store, even if that order is not useful. So the name membar_enter currently lives in a bad place where it means either of two things -- r/rw or w/rw. With this change, we deprecate membar_enter/exit, introduce membar_acquire/release as better names for the useful pair (r/rw and rw/w), and make sure the implementation of membar_enter guarantees both what was documented _and_ what was implemented, making it an alias for membar_sync. While here, rework all of the membar_* definitions and aliases. The new logic follows a rule to make it easier to audit: membar_X is defined as an alias for membar_Y iff membar_X is guaranteed by membar_Y. The `no stronger than' relation is (the transitive closure of): - membar_consumer (r/r) is guaranteed by membar_acquire (r/rw) - membar_producer (w/w) is guaranteed by membar_release (rw/w) - membar_acquire (r/rw) is guaranteed by membar_sync (rw/rw) - membar_release (rw/w) is guaranteed by membar_sync (rw/rw) And, for the deprecated membars: - membar_enter (whether r/rw, w/rw, or rw/rw) is guaranteed by membar_sync (rw/rw) - membar_exit (rw/w) is guaranteed by membar_release (rw/w) (membar_exit is identical to membar_release, but the name is deprecated.) Finally, while here, annotate some of the instructions with their semantics. For powerpc, leave an essay with citations on the unfortunate but -- as far as I can tell -- necessary decision to use lwsync, not isync, for membar_acquire and membar_consumer. Also add membar(3) and atomic(3) man page links. 1.4 Sat Apr 09 23:32:52 GMT 2022 riastradh Introduce membar_acquire/release. Deprecate membar_enter/exit. The names membar_enter/exit were unclear, and the documentation of membar_enter has disagreed with the implementations on sparc, powerpc, and even x86(!) for the entire time it has been in NetBSD. The terms `acquire' and `release' are ubiquitous in the literature today, and have been adopted in the C and C++ standards to mean load-before-load/store and load/store-before-store, respectively, which are exactly the orderings required by acquiring and releasing a mutex, as well as other useful applications like decrementing a reference count and then freeing the underlying object if it went to zero. Originally I proposed changing one word in the documentation for membar_enter to make it load-before-load/store instead of store-before-load/store, i.e., to make it an acquire barrier. I proposed this on the grounds that (a) all implementations guarantee load-before-load/store, (b) some implementations fail to guarantee store-before-load/store, and (c) all uses in-tree assume load-before-load/store. I verified parts (a) and (b) (except, for (a), powerpc didn't even guarantee load-before-load/store -- isync isn't necessarily enough; need lwsync in general -- but it _almost_ did, and it certainly didn't guarantee store-before-load/store). Part (c) might not be correct, however: under the mistaken assumption that atomic-r/m/w then membar-w/rw is equivalent to atomic-r/m/w then membar-r/rw, I only audited the cases of membar_enter that _aren't_ immediately after an atomic-r/m/w. All of those cases assume load-before-load/store. But my assumption was wrong -- there are cases of atomic-r/m/w then membar-w/rw that would be broken by changing to atomic-r/m/w then membar-r/rw: https://mail-index.netbsd.org/tech-kern/2022/03/29/msg028044.html Furthermore, the name membar_enter has been adopted in other places like OpenBSD where it actually does follow the documentation and guarantee store-before-load/store, even if that order is not useful. So the name membar_enter currently lives in a bad place where it means either of two things -- r/rw or w/rw. With this change, we deprecate membar_enter/exit, introduce membar_acquire/release as better names for the useful pair (r/rw and rw/w), and make sure the implementation of membar_enter guarantees both what was documented _and_ what was implemented, making it an alias for membar_sync. While here, rework all of the membar_* definitions and aliases. The new logic follows a rule to make it easier to audit: membar_X is defined as an alias for membar_Y iff membar_X is guaranteed by membar_Y. The `no stronger than' relation is (the transitive closure of): - membar_consumer (r/r) is guaranteed by membar_acquire (r/rw) - membar_producer (w/w) is guaranteed by membar_release (rw/w) - membar_acquire (r/rw) is guaranteed by membar_sync (rw/rw) - membar_release (rw/w) is guaranteed by membar_sync (rw/rw) And, for the deprecated membars: - membar_enter (whether r/rw, w/rw, or rw/rw) is guaranteed by membar_sync (rw/rw) - membar_exit (rw/w) is guaranteed by membar_release (rw/w) (membar_exit is identical to membar_release, but the name is deprecated.) Finally, while here, annotate some of the instructions with their semantics. For powerpc, leave an essay with citations on the unfortunate but -- as far as I can tell -- necessary decision to use lwsync, not isync, for membar_acquire and membar_consumer. Also add membar(3) and atomic(3) man page links. |
| /src/sys/modules/compat_netbsd32_90/ | ||
| H A D | Makefile | 1.1 Sun Sep 22 22:59:40 GMT 2019 christos branches: 1.1.8; Add a new member to struct vfsstat and grow the unused members The new member is caled f_mntfromlabel and it is the dkw_wname of the corresponding wedge. This is now used by df -W to display the mountpoint name as NAME= |
| /src/sys/arch/sparc64/include/ | ||
| H A D | wchar_limits.h | 1.1 Sat May 08 21:51:51 GMT 2004 kleink branches: 1.1.2; Factor out W{CHAR,INT}_{MAX,MIN} into their own header file. |
| /src/sys/arch/sun2/include/ | ||
| H A D | wchar_limits.h | 1.1 Sat May 08 21:51:51 GMT 2004 kleink branches: 1.1.2; Factor out W{CHAR,INT}_{MAX,MIN} into their own header file. |
| /src/sys/arch/sun3/include/ | ||
| H A D | wchar_limits.h | 1.1 Sat May 08 21:51:51 GMT 2004 kleink branches: 1.1.2; Factor out W{CHAR,INT}_{MAX,MIN} into their own header file. |
| /src/sys/arch/x68k/include/ | ||
| H A D | wchar_limits.h | 1.1 Sat May 08 21:51:51 GMT 2004 kleink branches: 1.1.2; Factor out W{CHAR,INT}_{MAX,MIN} into their own header file. |
| /src/sys/arch/atari/include/ | ||
| H A D | wchar_limits.h | 1.1 Sat May 08 21:51:47 GMT 2004 kleink branches: 1.1.2; Factor out W{CHAR,INT}_{MAX,MIN} into their own header file. |
| /src/sys/arch/bebox/include/ | ||
| H A D | wchar_limits.h | 1.1 Sat May 08 21:51:47 GMT 2004 kleink branches: 1.1.2; Factor out W{CHAR,INT}_{MAX,MIN} into their own header file. |
| /src/sys/arch/cats/include/ | ||
| H A D | wchar_limits.h | 1.1 Sat May 08 21:51:47 GMT 2004 kleink branches: 1.1.2; Factor out W{CHAR,INT}_{MAX,MIN} into their own header file. |
| /src/sys/arch/cesfic/include/ | ||
| H A D | wchar_limits.h | 1.1 Sat May 08 21:51:47 GMT 2004 kleink branches: 1.1.2; Factor out W{CHAR,INT}_{MAX,MIN} into their own header file. |
| /src/sys/arch/cobalt/include/ | ||
| H A D | wchar_limits.h | 1.1 Sat May 08 21:51:48 GMT 2004 kleink branches: 1.1.2; Factor out W{CHAR,INT}_{MAX,MIN} into their own header file. |
| /src/sys/arch/dreamcast/include/ | ||
| H A D | wchar_limits.h | 1.1 Sat May 08 21:51:48 GMT 2004 kleink branches: 1.1.2; Factor out W{CHAR,INT}_{MAX,MIN} into their own header file. |
| /src/sys/compat/common/ | ||
| H A D | vfs_syscalls_90.c | 1.1 Sun Sep 22 22:59:38 GMT 2019 christos branches: 1.1.8; Add a new member to struct vfsstat and grow the unused members The new member is caled f_mntfromlabel and it is the dkw_wname of the corresponding wedge. This is now used by df -W to display the mountpoint name as NAME= |
| /src/sys/arch/prep/include/ | ||
| H A D | wchar_limits.h | 1.1 Sat May 08 21:51:50 GMT 2004 kleink branches: 1.1.2; Factor out W{CHAR,INT}_{MAX,MIN} into their own header file. |
| /src/sys/arch/sandpoint/include/ | ||
| H A D | wchar_limits.h | 1.1 Sat May 08 21:51:50 GMT 2004 kleink branches: 1.1.2; Factor out W{CHAR,INT}_{MAX,MIN} into their own header file. |
Completed in 61 milliseconds