Re: [gpsd-dev] test failures with shmget

[Gary E. Miller]

Yo Greg!

On Tue, 03 Feb 2015 16:06:03 -0500
Greg Troxel <address@hidden> wrote:

> > The ntpd guys are rethinking the SHM permissions, so any gpsd fix
> > will be very short lived.  OTOH, this does seem to be a new
> > regression.
> 
> Well, I guess if it was just what NTP needed, that would be one thing,
> but it seems to leave more and more and run out.  I'm not really sure,
> but I've found 6-8 each on machines where the tests failed.

ntpd allows for up to about 200 SHM's.

If you convert the hex keys to ASCII, you get NTP0, NTP1...

There are two for each gps, even for the PVT time, odd for the PPS time.

I do see a real oddity after trying to run here:

robin ~ # ipcs
[...]
------ Shared Memory Segments --------
key        shmid      owner      perms      bytes      nattch     status      
0x4e545030 0          root       600        96         2                       
0x4e545031 32769      root       600        96         2                       
0x4e545032 65538      root       666        96         1                       
0x4e545033 98307      root       666        96         1                       
0x47505344 131076     root       666        8248       0                       
0x4e545034 163845     root       666        96         1                       
0x4e545035 196614     root       666        96         1                       
0x4e545036 229383     root       666        96         1                       
0x4e545037 262152     root       666        96         1                       

Notice NTP4 (0x47505344)  has a size of 8248.  They should all be 96!

Clearly a bug.

RGDS
GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97701
        address@hidden  Tel:+1(541)382-8588

pgp7nVfDS8Fu7.pgp
Description: OpenPGP digital signature