Re: [PATCH 4.14 000/101] 4.14.92-stable review

[Guenter Roeck]

On Wed, Jan 09, 2019 at 03:46:58PM +0100, Greg Kroah-Hartman wrote:
> On Wed, Jan 09, 2019 at 12:40:11PM +0100, Jinpu Wang wrote:
> > Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> ä2019å1æ8æåä äå1:28åéï
> > >
> > > On Mon, Jan 07, 2019 at 08:59:12PM -0800, Guenter Roeck wrote:
> > >
> > > Should all now be fixed up.
> > >
> > > I hope, this set of releases has been a pain...
> > >
> > > greg k-h
> > 
> > I had build failure below, looks related to 0fe69354b888 ("crypto:
> > testmgr - add AES-CFB tests")
...
> > make[4]: *** [crypto/testmgr.o] Error 1
> > Makefile:1045: recipe for target 'crypto' failed
> > make[3]: *** [crypto] Error 2
> 
> You are right, that totally does not build.  Odd that no one else's
> tests caught this :(
> 
The failure is only seen with CONFIG_CRYPTO_MANAGER_DISABLE_TESTS=n,
but CONFIG_CRYPTO_MANAGER_DISABLE_TESTS=y by default, meaning it is
explicitly disabled for all test builds. Out of the probably thousands
of default configurations, only two s390 configurations enable those
tests (ie explicitly set CONFIG_CRYPTO_MANAGER_DISABLE_TESTS=n).
So that is not at all surprising.

Question should be why the responsible developers pretty much ensured
that this code is never even build tested.

Guenter