On Wed, Nov 4, 2020 at 1:39 AM Alejandro Colomar <colomar.6.4.3@xxxxxxxxx> wrote: > > Maybe I'd also add a paragraph in VERSIONS, saying there was an old > version with a different prototype (and maybe show it there). > > What do you think about it? Hi Alex, I don't mind your re-doing my patch with the typo fix, updated examples, and the addition to versions. Please let me know if you would like to do that. If not, I'll send a new patch with these changes. Thanks, Peter > > Thanks, > > Alex > > On 2020-11-03 22:01, Alejandro Colomar wrote: > > Hi Peter, > > > > It looks very good! Just one typo: > > > > [[ > > .BI "int membarrier(int " cmd ", unsigned int " flags ", int " cpu_id );" > > ]] > > should be: > > [[ > > .BI "int membarrier(int " cmd ", unsigned int " flags ", int " cpu_id ); > > ]] > > > > Note the removal of the last '"', which was unpaired. > > > > > > Thanks, > > > > Alex > > > > On 2020-11-03 20:04, Peter Oskolkov wrote: > >> Linux kernel commit 2a36ab717e8fe678d98f81c14a0b124712719840 > >> (part of 5.10 release) changed sys_membarrier prototype/parameters > >> and added two new commands. This manpages patch reflects these > >> changes, by mostly copying comments from the kernel patch > >> into the manpage (I was also the author of the kernel change). > >> --- > >> man2/membarrier.2 | 41 +++++++++++++++++++++++++++++++++++++++-- > >> 1 file changed, 39 insertions(+), 2 deletions(-) > >> > >> diff --git a/man2/membarrier.2 b/man2/membarrier.2 > >> index 3064b2d2e..c95e97cb6 100644 > >> --- a/man2/membarrier.2 > >> +++ b/man2/membarrier.2 > >> @@ -30,7 +30,7 @@ membarrier \- issue memory barriers on a set of threads > >> .PP > >> .B #include <linux/membarrier.h> > >> .PP > >> -.BI "int membarrier(int " cmd ", int " flags ");" > >> +.BI "int membarrier(int " cmd ", unsigned int " flags ", int " cpu_id > >> );" > >> .fi > >> .PP > >> .IR Note : > >> @@ -165,6 +165,29 @@ core command prior to using it. > >> Register the process's intent to use > >> .BR MEMBARRIER_CMD_PRIVATE_EXPEDITED_SYNC_CORE . > >> .TP > >> +.BR MEMBARRIER_CMD_PRIVATE_EXPEDITED_RSEQ " (since Linux 5.10)" > >> +Ensure the caller thread, upon return from system call, that all its > >> +running thread siblings have any currently running rseq critical > >> sections > >> +restarted if > >> +.I flags > >> +parameter is 0; if > >> +.I flags > >> +parameter is > >> +.BR MEMBARRIER_CMD_FLAG_CPU , > >> +then this operation is performed only on CPU indicated by > >> +.I cpu_id . > >> +This guarantee is provided only for threads in > >> +the same process as the calling thread. > >> +.IP > >> +RSEQ membarrier is only available in the "private expedited" form. > >> +.IP > >> +A process must register its intent to use the private expedited rseq > >> +command prior to using it. > >> +.TP > >> +.BR MEMBARRIER_CMD_REGISTER_PRIVATE_EXPEDITED_RSEQ " (since Linux 5.10)" > >> +Register the process's intent to use > >> +.BR MEMBARRIER_CMD_PRIVATE_EXPEDITED_RSEQ . > >> +.TP > >> .BR MEMBARRIER_CMD_SHARED " (since Linux 4.3)" > >> This is an alias for > >> .BR MEMBARRIER_CMD_GLOBAL > >> @@ -172,7 +195,21 @@ that exists for header backward compatibility. > >> .PP > >> The > >> .I flags > >> -argument is currently unused and must be specified as 0. > >> +argument must be specified as 0 unless the command is > >> +.BR MEMBARRIER_CMD_PRIVATE_EXPEDITED_RSEQ , > >> +in which case > >> +.I flags > >> +can be either 0 or > >> +.BR MEMBARRIER_CMD_FLAG_CPU . > >> +.PP > >> +The > >> +.I cpu_id > >> +argument is ignored unless > >> +.I flags > >> +is > >> +.BR MEMBARRIER_CMD_FLAG_CPU , > >> +in which case it must specify the CPU targeted by this membarrier > >> +command. > >> .PP > >> All memory accesses performed in program order from each targeted > >> thread > >> are guaranteed to be ordered with respect to > >>
