[LTP] [PATCH v5 4/4] umip_basic_test.c: improve kconfig verification to avoid umip wrong abort case

[Xu, Pengfei]

Hi Petr,
  Seems LINUX_VERSION_CODE way it not suitable when test platform is not compiled platform.
  Need to use " if ((tst_kvercmp(5, 5, 0)) >= 0)" way.

Thanks!
BR
Pengfei
+86 021 61164364   



-----Original Message-----
From: Pengfei Xu <pengfei.xu@intel.com> 
Sent: Tuesday, May 26, 2020 18:37
To: Petr Vorel <pvorel@suse.cz>
Cc: ltp <ltp@lists.linux.it>; Neri, Ricardo <ricardo.neri@intel.com>; Su, Heng <heng.su@intel.com>; Kasten, Robert A <robert.a.kasten@intel.com>; Cyril Hrubis <chrubis@suse.cz>; Jan Stancek <jstancek@redhat.com>; Li Wang <liwang@redhat.com>
Subject: Re: [LTP] [PATCH v5 4/4] umip_basic_test.c: improve kconfig verification to avoid umip wrong abort case

Hi Petr,

On 2020-05-26 at 12:11:33 +0200, Petr Vorel wrote:
> Hi Pengfei,
> 
> > > But it looks like Cyril is not against the implementation, it just 
> > > needs to be
> > > fixed:
> > > https://patchwork.ozlabs.org/comment/2352151/
> 
> > You are right, actually it could be worked as my suggest way:
> > "CONFIG_A|CONFIG_B=Y".
> > I tried to use Cyril's advice "CONFIG_A=X|CONFIG_B=Y" way, which 
> > will add more code complexity, so I just want to solve the problem I 
> > am currently facing.
> > If we really need the "CONFIG_A=X|CONFIG_B=Y" function, which cannot 
> > be satisfied by "CONFIG_A|CONFIG_B=Y" function in the future, then 
> > we could add this function I think.
> > Thanks for your considering.
> 
> I'd also think that we need "CONFIG_A=X|CONFIG_B=Y", because 
> "CONFIG_A|CONFIG_B=Y" is ambiguous (we support both CONFIG_FOO and 
> CONFIG_FOO=bar and this must stay even with |).
> 
> Will you send a patch for that or shell I fix it with 
> LINUX_VERSION_CODE < KERNEL_VERSION for now?

  Ok, thanks for your LINUX_VERSION way to fix this issue.
  Thanks!
  BR.

> 
> Kind regards,
> Petr