gpsd-dev
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [gpsd-dev] scons check fail gpsd3.14 ubuntu 14


From: Hal Murray
Subject: Re: [gpsd-dev] scons check fail gpsd3.14 ubuntu 14
Date: Wed, 18 Mar 2015 02:03:14 -0700

>> If chrpath is no, then make check should error out.
> An interesting idea, but we could be on a platform where chrpath is not
> available.  Do we really want to force the user to static-link in that case?

I poked around a bit more.

Here is the condensed version from my man page for ld.so/ld-linux-so:
If  RUNPATH doesn't exist:
  search RPATH, then LD_LIBRARY_PATH, then normal stuff.
else
  search LD_LIBRARY_PATH, then RUNPATH, then normal stuff.

It also says that RPATH is deprecated.  It doesn't say why, but this might be 
the reason.

(It calls them DT_RPATH and DT_RUNPATH.)

I don't know my way around scons/SConstruct well enough to fix it.
I added "chrpath -c ./gpsd" to the top of regress-driver.
That now works on a chrpath=no setup that used to do crazy things.

ldd can be used to printout which libraries will be used.
It processes LD_LIBRARY_PATH
A bit of hacking confirms the search order descriptions on the ld.so man page.

I think you have to add ",--enable-new-dtags" to the -Wl,rpath=xxx line to 
make the linker produce RUNPATH rather than RPATH.

I don't know how many systems do/don't support RUNPATH.  I think it's been 
around for a long time.  It's supported on a very old linux box as well as 
current NetBSD and FreeBSD.



-- 
These are my opinions.  I hate spam.






reply via email to

[Prev in Thread] Current Thread [Next in Thread]