This is following a failure to build "arch/x86/xen/enlighten.c" in
build-i386-pvops jobs with linux 968f35f4ab1c ("Merge tag
'v6.7-rc3-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6")
in linux-linus branch.

Suggested-by: Juergen Gross <jgross@suse.com>
Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>
---
 mfi-common | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/mfi-common b/mfi-common
index 1a131c27..d8d167fd 100644
--- a/mfi-common
+++ b/mfi-common
@@ -319,6 +319,11 @@ create_build_jobs () {
         revision_linux=${REVISION_LINUX_ARM:-${DEFAULT_REVISION_LINUX_ARM}}
       "
       ;;
+    i386)
+      pvops_kconfig_overrides="
+        kconfig_override_y=CONFIG_X86_GENERIC
+      "
+      ;;
     esac
 
     case "$arch" in
-- 
Anthony PERARD

On Fri, Dec 08, 2023 at 04:02:25PM +0000, Anthony PERARD wrote:
> This is following a failure to build "arch/x86/xen/enlighten.c" in
> build-i386-pvops jobs with linux 968f35f4ab1c ("Merge tag
> 'v6.7-rc3-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6")
> in linux-linus branch.

Could we expand a bit on why CONFIG_X86_GENERIC is required?  It would
seem to me that option should have been selected by default already?

Thanks, Roger.

On 12.12.23 09:38, Roger Pau Monné wrote:
> On Fri, Dec 08, 2023 at 04:02:25PM +0000, Anthony PERARD wrote:
>> This is following a failure to build "arch/x86/xen/enlighten.c" in
>> build-i386-pvops jobs with linux 968f35f4ab1c ("Merge tag
>> 'v6.7-rc3-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6")
>> in linux-linus branch.
> 
> Could we expand a bit on why CONFIG_X86_GENERIC is required?  It would
> seem to me that option should have been selected by default already?

No, the kernel's default for 32-bit x86 builds is still using the M686.

I guess this is for historical reasons.

BTW, there is a kernel patch pending letting CONFIG_XEN depend on more
recent cpu types for 32-bit builds. I'm planning to take that patch
after OSStest has been changed to no longer use M686.


Juergen

On Tue, Dec 12, 2023 at 09:48:46AM +0100, Juergen Gross wrote:
> On 12.12.23 09:38, Roger Pau Monné wrote:
> > On Fri, Dec 08, 2023 at 04:02:25PM +0000, Anthony PERARD wrote:
> > > This is following a failure to build "arch/x86/xen/enlighten.c" in
> > > build-i386-pvops jobs with linux 968f35f4ab1c ("Merge tag
> > > 'v6.7-rc3-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6")
> > > in linux-linus branch.
> > 
> > Could we expand a bit on why CONFIG_X86_GENERIC is required?  It would
> > seem to me that option should have been selected by default already?
> 
> No, the kernel's default for 32-bit x86 builds is still using the M686.
> 
> I guess this is for historical reasons.

If I add the following to the patch description, taken from Juergen
patch, would that be ok?

    Today the cpu type for a 32-bit x86 kernel is not specified in the
    used kernel config, resulting in the M686 to be used.

    Instead of using the M686 which isn't even a 64-bit cpu (thus not
    capable to run a Xen guest), use the X86_GENERIC variant which is
    more appropriate.

ref: https://lore.kernel.org/all/20231207080953.27976-1-jgross@suse.com/

I'm not sure what you mean Juergen with "not 64bit cpu -> can't run xen
guest", maybe it's because we can only run Xen on 64-bit cpus nowadays,
so if we can't run Xen, there's no Xen guest?

As for X86_GENERIC, I think I've read the description which says
something like distro should enable it, so I guess 32-bit distro would
usually have the config enabled, so it make sense that we test with it.

Cheers,

-- 
Anthony PERARD

On 12.12.23 12:56, Anthony PERARD wrote:
> On Tue, Dec 12, 2023 at 09:48:46AM +0100, Juergen Gross wrote:
>> On 12.12.23 09:38, Roger Pau Monné wrote:
>>> On Fri, Dec 08, 2023 at 04:02:25PM +0000, Anthony PERARD wrote:
>>>> This is following a failure to build "arch/x86/xen/enlighten.c" in
>>>> build-i386-pvops jobs with linux 968f35f4ab1c ("Merge tag
>>>> 'v6.7-rc3-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6")
>>>> in linux-linus branch.
>>>
>>> Could we expand a bit on why CONFIG_X86_GENERIC is required?  It would
>>> seem to me that option should have been selected by default already?
>>
>> No, the kernel's default for 32-bit x86 builds is still using the M686.
>>
>> I guess this is for historical reasons.
> 
> If I add the following to the patch description, taken from Juergen
> patch, would that be ok?
> 
>      Today the cpu type for a 32-bit x86 kernel is not specified in the
>      used kernel config, resulting in the M686 to be used.
> 
>      Instead of using the M686 which isn't even a 64-bit cpu (thus not
>      capable to run a Xen guest), use the X86_GENERIC variant which is
>      more appropriate.
> 
> ref: https://lore.kernel.org/all/20231207080953.27976-1-jgross@suse.com/
> 
> I'm not sure what you mean Juergen with "not 64bit cpu -> can't run xen
> guest", maybe it's because we can only run Xen on 64-bit cpus nowadays,
> so if we can't run Xen, there's no Xen guest?

Correct.


Juergen

On 08.12.23 17:02, Anthony PERARD wrote:
> This is following a failure to build "arch/x86/xen/enlighten.c" in
> build-i386-pvops jobs with linux 968f35f4ab1c ("Merge tag
> 'v6.7-rc3-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6")
> in linux-linus branch.
> 
> Suggested-by: Juergen Gross <jgross@suse.com>
> Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

Thanks, much appreciated.


Juergen (not knowing enough of OSStest to give a "R-b:" for the patch)

> ---
>   mfi-common | 5 +++++
>   1 file changed, 5 insertions(+)
> 
> diff --git a/mfi-common b/mfi-common
> index 1a131c27..d8d167fd 100644
> --- a/mfi-common
> +++ b/mfi-common
> @@ -319,6 +319,11 @@ create_build_jobs () {
>           revision_linux=${REVISION_LINUX_ARM:-${DEFAULT_REVISION_LINUX_ARM}}
>         "
>         ;;
> +    i386)
> +      pvops_kconfig_overrides="
> +        kconfig_override_y=CONFIG_X86_GENERIC
> +      "
> +      ;;
>       esac
>   
>       case "$arch" in