Re: [ptxdist] [ANNOUNCE] OSELAS.Toolchain() 2016.06.0 released

[Clemens Gruber <clemens.gruber@pqgruber.com>]

Hi,

On Mon, Jul 04, 2016 at 08:35:44AM +0200, Michael Olbrich wrote:
> Updating md5sums with '--update-md5' when updating toolchain components
> works for me. I have no idea why this doesn't work for you.

You explained this in the last paragraph. It's only affecting symbols
like PTXCONF_GDBSERVER_MD5, which should not exist at all.
So that's fine. For PTXCONF_GDB_MD5, --update-md5 works for me too.

What's odd though is that the toolchain build process expects not only
the PTXCONF_<PKG>_MD5 lines but also lines where PKG is CROSS_GCC or
GLIBC_FIRST, GLIBC_HEADERS or GLIBC_CRT.
Only if I add those it continues. And next time I change something via
ptxdist menu, they get cleaned up again (because they should not exist
in the first place).

But then, why are these symbols necessary for the build?
Shouldn't cross-gcc just use PTXCONF_GCC_MD5 and glibc-first just use
PTXCONF_GLIBC_MD5 instead of PTXCONF_GLIBC_FIRST_MD5?

> There is no PTXCONF_GDBSERVER_MD5. In this case gdbserver reuses
> PTXCONF_GDB_MD5, which is the reason why '--update-md5' doesn't work here.

I think it's this very process of reusing kconfig symbols like
PTXCONF_GDB_MD5 during the gdbserver build, which does not work as
expected.

Same behavior as with the toolchain builds. When I change one of the
versions through ptxdist menu and do a clean rebuild, this occurs.

Could this be some kind of fallback if the MD5 hash in PTXCONF_GDB_MD5 
is not correct after the version update, trying other symbol names like
PTXCONF_GDBSERVER_MD5?

Thanks for your help!

Clemens

_______________________________________________
ptxdist mailing list
ptxdist@pengutronix.de