On 9/15/18 7:34 PM, Simon Marchi wrote:
On 2018-09-12 7:37 p.m., John Baldwin wrote:
This should include most of the suggested documentation fixes from
the
first series. It also adds an additional patch that attempts to tidy
up
some of the other "info proc" documentation. One change I haven't
made
(wasn't sure if it was still desired) was if we wanted to replace the
specific annotations on individual 'info proc' subcommands about
which
OS's supported those commands with a single, more general statement
that
commands may only be supported on a subset of systems supported by
GDB.
I've moved more of the shared code for generating the 'info proc
files'
output to fbsd-tdep.c.
One open question still from the first series is if GDB can assume
the
presence of routines like 'inet_ntoa' and 'inet_ntop' for formatting
IPv4 and IPv6 addresses. It seems GDB does assume the presence of
newer routines (e.g. getaddrinfo()) on POSIX systems in other places
(e.g. ser-tcp.c), though in those places we use different APIs for
Win32.
There seems to be a gnulib module for that:
https://www.gnu.org/software/gnulib/manual/html_node/inet_005fntop.html
There is a page for inet_ntoa too:
https://www.gnu.org/software/gnulib/manual/html_node/inet_005fntoa.html
but there is no gnulib module for it. Probably because we can always
use inet_ntop instead?
Yes, inet_ntop is sufficient for both. I had grepp'ed for 'ntoa' and
'ntop' in gdb/gnulib and didn't find any matches, so I assumed that
meant there wasn't a module. Is gdb/gnulib a subset of the actual
gnulib
and new bits are imported on demand?
Hmm, reading update-gnulib.sh, it seems so. I'll look at what is
involved
in doing that.