On Sun, Dec 03, 2017 at 12:51:16PM +0000, Sami Kerola wrote: > Add more information when and why one might want to use various options. > > Signed-off-by: Sami Kerola <kerolasa@xxxxxx> > --- > sys-utils/setarch.8 | 51 +++++++++++++++++++++++++++++++++++++++++---------- > 1 file changed, 41 insertions(+), 10 deletions(-) > > diff --git a/sys-utils/setarch.8 b/sys-utils/setarch.8 > index 73874c8b6..ce7ba97ac 100644 > --- a/sys-utils/setarch.8 > +++ b/sys-utils/setarch.8 [...] > .TP > \fB\-\-4gb\fR > -Ignored (for backward compatibility only). > +Ignored for backward compatibility only. The new wording is more confusing than the old one. [...] > .TP > \fB\-F\fR, \fB\-\-fdpic\-funcptrs\fR > -Userspace function pointers point to descriptors (turns on FDPIC_FUNCPTRS). > +User-space function pointers to signal handlers point. Use when kernel does > +not have MMU. Supported on ARM, Blackfin, Fujitsu FR-V, and SuperH (as of > +2017-12-02). Turns on FDPIC_FUNCPTRS. The new wording is more confusing than the old one. [...] > .TP > \fB\-T\fR, \fB\-\-sticky\-timeouts\fR > +This makes > +.BR select (2), > +.BR pselect (2), > +and > +.BR ppoll (2) > +system calls preserve the timeout value instead of storing the unslept time > +when interrupted by a signal handler. Use when > +.I program > +depends on this behavior, and the source code is not available to be fixed. > Turns on STICKY_TIMEOUTS. The use of word "fixed" is not appropriate here. According to select(2), "On Linux, select() modifies timeout to reflect the amount of time not slept; most other implementations do not do this. (POSIX.1 permits either behavior.) This causes problems both when Linux code which reads timeout is ported to other operating systems, and when code is ported to Linux that reuses a struct timeval for multiple select()s in a loop without reinitializing it. Consider timeout to be undefined after select() returns." -- ldv
Attachment:
signature.asc
Description: PGP signature
