On 23/08/2017 10:06, Thomas Huth wrote:
> On 23.08.2017 10:00, David Hildenbrand wrote:
>> On 23.08.2017 08:53, Christian Borntraeger wrote:
>>> KVM guests on s390 need a different page table layout than normal
>>> processes (2kb page table + 2kb page status extensions vs 2kb page table
>>> only). As of today this has to be enabled via the vm.allocate_pgste
>>> sysctl.
>>>
>>> Newer kernels (>= 4.12) on s390 check for an S390_PGSTE program header
>>> and enable the pgste page table extensions in that case. This makes the
>>> vm.allocate_pgste sysctl unnecessary. We enable this program header for
>>> the s390 system emulation (qemu-system-s390x) if we build on s390
>>> - for s390 system emulation
>>> - the linker supports --s390-pgste (binutils >= 2.29)
>>> - KVM is enabled
>>>
>>> This will allow distributions to disable the global vm.allocate_pgste
>>> sysctl, which will improve the page table allocation for non KVM
>>> processes as only 2kb chunks are necessary.
>>>
>>> Cc: Christian Ehrhardt <christian.ehrha...@canonical.com>
>>> Cc: Alexander Graf <ag...@suse.de>
>>> Cc: Dan Horak <dho...@redhat.com>
>>> Cc: David Hildenbrand <da...@redhat.com>
>>> Signed-off-by: Christian Borntraeger <borntrae...@de.ibm.com>
>>> Acked-by: Janosch Frank <fran...@linux.vnet.ibm.com>
>>> ---
>>> V1->V2:
>>> - provide ld_has function
>>> - use ld_has to replace some open coded variants
>>> - check target arch and arch for s390
>>> - check for s390x before calling the linker
>>>
>>> configure | 21 ++++++++++++++++++++-
>>> 1 file changed, 20 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/configure b/configure
>>> index dd73cce..0b68b37 100755
>>> --- a/configure
>>> +++ b/configure
>>> @@ -240,6 +240,11 @@ supported_target() {
>>> return 1
>>> }
>>>
>>> +
>>
>> I'd drop this new line
>>
>>> +ld_has() {
>>> + $ld --help 2>/dev/null | grep ".$1" >/dev/null 2>&1
>>> +}
>>> +
>>> # default parameters
>>> source_path=$(dirname "$0")
>>> cpu=""
>>> @@ -5043,7 +5048,7 @@ fi
>>> # Use ASLR, no-SEH and DEP if available
>>> if test "$mingw32" = "yes" ; then
>>> for flag in --dynamicbase --no-seh --nxcompat; do
>>> - if $ld --help 2>/dev/null | grep ".$flag" >/dev/null 2>/dev/null ;
>>> then
>>> + if ld_had $flag ; then
>>> LDFLAGS="-Wl,$flag $LDFLAGS"
>>> fi
>>> done
>>> @@ -6522,6 +6527,20 @@ if test "$target_linux_user" = "yes" -o
>>> "$target_bsd_user" = "yes" ; then
>>> ldflags="$ldflags $textseg_ldflags"
>>> fi
>>>
>>> +# Newer kernels on s390 check for an S390_PGSTE program header and
>>> +# enable the pgste page table extensions in that case. This makes
>>> +# the vm.allocate_pgste sysctl unnecessary. We enable this program
>>> +# header if
>>> +# - we build on s390x
>>> +# - we build the system emulation for s390x (qemu-system-s390x)
>>> +# - KVM is enabled
>>> +# - the linker support --s390-pgste
>>> +if test "$TARGET_ARCH" = "s390x" -a "$target_softmmu" = "yes" -a "$ARCH"
>>> = "s390x" -a "$kvm" = "yes"; then
>>
>> Wonder if the "$ARCH" check is really necessary: TARGET_ARCH=s390x with
>> kvm=yes should only build on s390x.
>
> Isn't kvm=yes and TARGET_ARCH=s390x also possible on a x86 host, where
> only the x86_64 target is built with CONFIG_KVM=y, but the s390x target
> with CONFIG_KVM=n ?
Yes. You could use
if test "$ARCH" = "s390x" && supported_kvm_target $target; then
...
fi
Or, in the existing "if supported_kvm_target $target" conditional, add
if test "$ARCH" = s390x && ld_has --s390-pgste; then
...
fi
Paolo
