The implementation of the pflocal server is in the pflocal directory, and uses ?libpipe (shared code with the named pipe implementation).

Open Issues


IRC, freenode, #hurd, 2013-09-19

<gnu_srs> Hi, is SO_REUSEADDR supported at all on Hurd? I can only find two
<gnu_srs> in libdde-linux26 and pfinet/linux-src, and the functionality
  seems to be unimplemented.
<pinotree> gnu_srs: pfinet supports it
<youpi> gnu_srs: grep talks about  pfinet/linux-src/net/core/sock.c:
<youpi> two times
<gnu_srs> Yes, and that is the implementation?
<gnu_srs> I wrote a test for AF_INET and it works, but not for AF_UNIX
  (maybe not so interesting case).
<pinotree> pflocal does not support it
<gnu_srs> Is that of interest at all?

IRC, freenode, #hurd, 2014-01-14

<braunr> sudo -s eats 100 cpu :/
<braunr> possibly because of pflocal
<braunr> only change on pflocal (notwithstanding the libraries) is
  "pflocal: improve the demuxer functions"
<braunr> teythoon: why did you change the order of the function calls in
  sock_demuxer ?
<youpi> for efficiency iirc
<braunr> yes, looks reasonable

IRC, freenode, #hurd, 2014-01-16

<braunr> i suspect the "improve the demuxer functions" changes may have
  hard-to-understand side effects 
<teythoon> yes, mostly being faster
<braunr> ah, the latest sudo has been fixed
<braunr> haha :)
<teythoon> ^^
<braunr> that one is easy to understand :)
<braunr> sudo was looping around calls to pflocal
<braunr> and exim crashed because of pfinet
<braunr> and those servers were only affected by these changes, other than
  the library ones which don't seem to apply at all
<braunr> but with sudo being fixed, i'm not sure it's relevant any more
<teythoon> i'd say being faster could easily cause hard-to-understand side
<braunr> ah, yes
<braunr> being faster isn't the side effect itself ;p
<braunr> nice, sudo was bugged on linux too, its behaviour matched its hurd
  version perfectly


Doesn't support fsysopts.