Re: PR/55557 CVS commit: src

[Andreas Gustafsson <gson%gson.org@localhost>]

The following reply was made to PR toolchain/55557; it has been noted by GNATS.

From: Andreas Gustafsson <gson%gson.org@localhost>
To: christos%NetBSD.org@localhost
Cc: gnats-bugs%netbsd.org@localhost
Subject: Re: PR/55557 CVS commit: src
Date: Tue, 21 Dec 2021 10:25:05 +0200

 Christos Zoulas wrote:
 >  PR/55557: Andreas Gustafsson" Introduce a new variable MKDEBUGKERNEL which
 >  as the name implies, includes a netbsd.gdb inside each kernel set:
 >  $ tar -tzvf kern-GENERIC.tgz
 >  -rwxr-xr-x  0 root   wheel 29398264 Dec 19 12:50 ./netbsd
 >  -rwxr-xr-x  0 root   wheel 208125880 Dec 19 12:50 ./netbsd.gdb
 
 I suppose this is a step in the right direction, and it could even be
 considered a fix for PR 50128, because it may now finally be possible
 to construct a custom build that can be installed automatically using
 anita and symbolically debugged using kgdb on a pair of qemu VMs for
 architectures other than i386.
 
 However, since MKDEBUGKERNEL is not enabled by default, and probably
 shouldn't be due to the set size and disk space requirement, it is not
 sufficient to fix the present PR, which is specifically about the
 official binary releases.  My preferred solution for those would be a
 separate debugkern-GENERIC set containing a netbsd.gdb, and possibly
 but not necessarily, corresponding debugkern-FOO sets for other
 kernels.
 
 It also does not solve the issue of kernel debugging on the testbeds
 (as in the recent case of PR 56554), because those should be testing
 configurations that are as close as possible to what we ship in the
 official binary releases.
 -- 
 Andreas Gustafsson, gson%gson.org@localhost