[PATCH v2 0/6] buildsys: Build faster (mostly tools and linux-user)

[PATCH v2 0/6] buildsys: Build faster (mostly tools and linux-user)

[Philippe Mathieu-Daudé]

In some configuration (linux-user, tools) we can ignore building
various objects (and the libfdt).

Tested with all the combinations of --[enable|disable]-tools,
--[enable|disable]-user and --[enable|disable]-system using the
following commands (suggested by Laurent Vivier in v1):

  $ mkdir build
  $ cd build
  $ for user in enable disable; do \
        for tools in enable disable; do \
            for system in enable disable; do \
                rm -fr build-$user-$system-$tools && \
                mkdir build-$user-$system-$tools && \
                    (cd build-$user-$system-$tools && \
                     ../../configure \
                                     --${user}-user \
                                     --${system}-system \
                                     --${tools}-tools \
                                     --disable-docs \
                    ); \
            done; \
        done; \
    done

Then building each of the 8 subdirectories on x86_64 and aarch64
hosts, running 'make check', and only on x86_64:
'make run-tcg-tests-x86_64-linux-user'.

All CI green:
https://gitlab.com/philmd/qemu/pipelines/110420332
https://travis-ci.org/philmd/qemu/builds/638781159
https://app.shippable.com/github/philmd/qemu/runs/587/summary/console

Since v1:
- no code change, improved commit description, added review tags
- added 2 new patches touching hw/core/ (remove reset.o from linux-user)

$ git backport-diff -u v1
Key:
[----] : patches are identical
[####] : number of functional differences between upstream/downstream patch
[down] : patch is downstream-only
The flags [FC] indicate (F)unctional and (C)ontextual differences, respectively

001/6:[----] [--] 'configure: Do not build libfdt if not required'
002/6:[----] [--] 'Makefile: Clarify all the codebase requires qom/ objects'
003/6:[----] [--] 'Makefile: Restrict system emulation and tools objects'
004/6:[----] [--] 'Makefile: Remove unhelpful comment'
005/6:[down] 'hw/core: Restrict reset handlers API to system-mode'
006/6:[down] 'hw/core/Makefile: Group generic objects versus system-mode objects'

Supersedes: <20200109153939.27173-1-philmd@redhat.com>

Philippe Mathieu-Daudé (6):
  configure: Do not build libfdt if not required
  Makefile: Clarify all the codebase requires qom/ objects
  Makefile: Restrict system emulation and tools objects
  Makefile: Remove unhelpful comment
  hw/core: Restrict reset handlers API to system-mode
  hw/core/Makefile: Group generic objects versus system-mode objects

 configure             |  2 ++
 Makefile.objs         | 31 ++++++++++---------------------
 hw/core/Makefile.objs | 31 ++++++++++++++++---------------
 3 files changed, 28 insertions(+), 36 deletions(-)

-- 
2.21.1

[PATCH v2 1/6] configure: Do not build libfdt if not required

[Philippe Mathieu-Daudé]

We only require libfdt for system emulation, in a small set
of architecture:

4077  # fdt support is mandatory for at least some target architectures,
4078  # so insist on it if we're building those system emulators.
4079  fdt_required=no
4080  for target in $target_list; do
4081    case $target in
4082      aarch64*-softmmu|arm*-softmmu|ppc*-softmmu|microblaze*-softmmu|mips64el-softmmu|riscv*-softmmu)
4083        fdt_required=yes

Do not build libfdt if we did not manually specified --enable-fdt,
or have one of the platforms that require it in our target list.

Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
---
v2: Improved description (thuth)
---
 configure | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/configure b/configure
index 557e4382ea..c67a7e73db 100755
--- a/configure
+++ b/configure
@@ -4095,6 +4095,8 @@ if test "$fdt_required" = "yes"; then
       "targets which need it (by specifying a cut down --target-list)."
   fi
   fdt=yes
+elif test "$fdt" != "yes" ; then
+  fdt=no
 fi
 
 if test "$fdt" != "no" ; then
-- 
2.21.1

Re: [PATCH v2 1/6] configure: Do not build libfdt if not required

[Alex Bennée]

Philippe Mathieu-Daudé <philmd@redhat.com> writes:

> We only require libfdt for system emulation, in a small set
> of architecture:
>
> 4077  # fdt support is mandatory for at least some target architectures,
> 4078  # so insist on it if we're building those system emulators.
> 4079  fdt_required=no
> 4080  for target in $target_list; do
> 4081    case $target in
> 4082      aarch64*-softmmu|arm*-softmmu|ppc*-softmmu|microblaze*-softmmu|mips64el-softmmu|riscv*-softmmu)
> 4083        fdt_required=yes
>
> Do not build libfdt if we did not manually specified --enable-fdt,
> or have one of the platforms that require it in our target list.
>
> Reviewed-by: Thomas Huth <thuth@redhat.com>
> Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
> Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

> ---
> v2: Improved description (thuth)
> ---
>  configure | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/configure b/configure
> index 557e4382ea..c67a7e73db 100755
> --- a/configure
> +++ b/configure
> @@ -4095,6 +4095,8 @@ if test "$fdt_required" = "yes"; then
>        "targets which need it (by specifying a cut down --target-list)."
>    fi
>    fdt=yes
> +elif test "$fdt" != "yes" ; then
> +  fdt=no
>  fi
>  
>  if test "$fdt" != "no" ; then


-- 
Alex Bennée

[PATCH v2 2/6] Makefile: Clarify all the codebase requires qom/ objects

[Philippe Mathieu-Daudé]

QEMU user-mode also requires the qom/ objects, it is not only
used by "system emulation and qemu-img". As we will use a big
if() block, move it upper in the "Common libraries for tools
and emulators" section.

Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
---
 Makefile.objs | 6 +-----
 1 file changed, 1 insertion(+), 5 deletions(-)

diff --git a/Makefile.objs b/Makefile.objs
index 7c1e50f9d6..5aae561984 100644
--- a/Makefile.objs
+++ b/Makefile.objs
@@ -2,6 +2,7 @@
 # Common libraries for tools and emulators
 stub-obj-y = stubs/
 util-obj-y = crypto/ util/ qobject/ qapi/
+qom-obj-y = qom/
 
 chardev-obj-y = chardev/
 
@@ -26,11 +27,6 @@ block-obj-m = block/
 
 crypto-obj-y = crypto/
 
-#######################################################################
-# qom-obj-y is code used by both qemu system emulation and qemu-img
-
-qom-obj-y = qom/
-
 #######################################################################
 # io-obj-y is code used by both qemu system emulation and qemu-img
 
-- 
2.21.1

Re: [PATCH v2 2/6] Makefile: Clarify all the codebase requires qom/ objects

[Alex Bennée]

Philippe Mathieu-Daudé <philmd@redhat.com> writes:

> QEMU user-mode also requires the qom/ objects, it is not only
> used by "system emulation and qemu-img". As we will use a big
> if() block, move it upper in the "Common libraries for tools
> and emulators" section.
>
> Reviewed-by: Thomas Huth <thuth@redhat.com>
> Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

> ---
>  Makefile.objs | 6 +-----
>  1 file changed, 1 insertion(+), 5 deletions(-)
>
> diff --git a/Makefile.objs b/Makefile.objs
> index 7c1e50f9d6..5aae561984 100644
> --- a/Makefile.objs
> +++ b/Makefile.objs
> @@ -2,6 +2,7 @@
>  # Common libraries for tools and emulators
>  stub-obj-y = stubs/
>  util-obj-y = crypto/ util/ qobject/ qapi/
> +qom-obj-y = qom/
>  
>  chardev-obj-y = chardev/
>  
> @@ -26,11 +27,6 @@ block-obj-m = block/
>  
>  crypto-obj-y = crypto/
>  
> -#######################################################################
> -# qom-obj-y is code used by both qemu system emulation and qemu-img
> -
> -qom-obj-y = qom/
> -
>  #######################################################################
>  # io-obj-y is code used by both qemu system emulation and qemu-img


-- 
Alex Bennée

[PATCH v2 3/6] Makefile: Restrict system emulation and tools objects

[Philippe Mathieu-Daudé]

Restrict all the system emulation and tools objects with a
Makefile IF (CONFIG_SOFTMMU OR CONFIG_TOOLS) check.

Using the same description over and over is not very helpful.
Use it once, just before the if() block.

Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
---
 Makefile.objs | 19 +++++++------------
 1 file changed, 7 insertions(+), 12 deletions(-)

diff --git a/Makefile.objs b/Makefile.objs
index 5aae561984..395dd1e670 100644
--- a/Makefile.objs
+++ b/Makefile.objs
@@ -4,16 +4,15 @@ stub-obj-y = stubs/
 util-obj-y = crypto/ util/ qobject/ qapi/
 qom-obj-y = qom/
 
+#######################################################################
+# code used by both qemu system emulation and qemu-img
+
+ifeq ($(call lor,$(CONFIG_SOFTMMU),$(CONFIG_TOOLS)),y)
+
 chardev-obj-y = chardev/
 
-#######################################################################
-# authz-obj-y is code used by both qemu system emulation and qemu-img
-
 authz-obj-y = authz/
 
-#######################################################################
-# block-obj-y is code used by both qemu system emulation and qemu-img
-
 block-obj-y = nbd/
 block-obj-y += block.o blockjob.o job.o
 block-obj-y += block/ scsi/
@@ -22,16 +21,12 @@ block-obj-$(CONFIG_REPLICATION) += replication.o
 
 block-obj-m = block/
 
-#######################################################################
-# crypto-obj-y is code used by both qemu system emulation and qemu-img
-
 crypto-obj-y = crypto/
 
-#######################################################################
-# io-obj-y is code used by both qemu system emulation and qemu-img
-
 io-obj-y = io/
 
+endif # CONFIG_SOFTMMU or CONFIG_TOOLS
+
 ######################################################################
 # Target independent part of system emulation. The long term path is to
 # suppress *all* target specific code in case of system emulation, i.e. a
-- 
2.21.1

Re: [PATCH v2 3/6] Makefile: Restrict system emulation and tools objects

[Alex Bennée]

Philippe Mathieu-Daudé <philmd@redhat.com> writes:

> Restrict all the system emulation and tools objects with a
> Makefile IF (CONFIG_SOFTMMU OR CONFIG_TOOLS) check.
>
> Using the same description over and over is not very helpful.
> Use it once, just before the if() block.
>
> Reviewed-by: Thomas Huth <thuth@redhat.com>
> Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

> ---
>  Makefile.objs | 19 +++++++------------
>  1 file changed, 7 insertions(+), 12 deletions(-)
>
> diff --git a/Makefile.objs b/Makefile.objs
> index 5aae561984..395dd1e670 100644
> --- a/Makefile.objs
> +++ b/Makefile.objs
> @@ -4,16 +4,15 @@ stub-obj-y = stubs/
>  util-obj-y = crypto/ util/ qobject/ qapi/
>  qom-obj-y = qom/
>  
> +#######################################################################
> +# code used by both qemu system emulation and qemu-img
> +
> +ifeq ($(call lor,$(CONFIG_SOFTMMU),$(CONFIG_TOOLS)),y)
> +
>  chardev-obj-y = chardev/
>  
> -#######################################################################
> -# authz-obj-y is code used by both qemu system emulation and qemu-img
> -
>  authz-obj-y = authz/
>  
> -#######################################################################
> -# block-obj-y is code used by both qemu system emulation and qemu-img
> -
>  block-obj-y = nbd/
>  block-obj-y += block.o blockjob.o job.o
>  block-obj-y += block/ scsi/
> @@ -22,16 +21,12 @@ block-obj-$(CONFIG_REPLICATION) += replication.o
>  
>  block-obj-m = block/
>  
> -#######################################################################
> -# crypto-obj-y is code used by both qemu system emulation and qemu-img
> -
>  crypto-obj-y = crypto/
>  
> -#######################################################################
> -# io-obj-y is code used by both qemu system emulation and qemu-img
> -
>  io-obj-y = io/
>  
> +endif # CONFIG_SOFTMMU or CONFIG_TOOLS
> +
>  ######################################################################
>  # Target independent part of system emulation. The long term path is to
>  # suppress *all* target specific code in case of system emulation, i.e. a


-- 
Alex Bennée

[PATCH v2 4/6] Makefile: Remove unhelpful comment

[Philippe Mathieu-Daudé]

It is pointless to keep qapi/ object separate from the other
common-objects. Drop the comment.

Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
---
 Makefile.objs | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/Makefile.objs b/Makefile.objs
index 395dd1e670..c6321d0465 100644
--- a/Makefile.objs
+++ b/Makefile.objs
@@ -71,11 +71,9 @@ qemu-seccomp.o-libs := $(SECCOMP_LIBS)
 
 common-obj-$(CONFIG_FDT) += device_tree.o
 
-######################################################################
-# qapi
-
 common-obj-y += qapi/
-endif
+
+endif # CONFIG_SOFTMMU
 
 #######################################################################
 # Target-independent parts used in system and user emulation
-- 
2.21.1

Re: [PATCH v2 4/6] Makefile: Remove unhelpful comment

[Alex Bennée]

Philippe Mathieu-Daudé <philmd@redhat.com> writes:

> It is pointless to keep qapi/ object separate from the other
> common-objects. Drop the comment.
>
> Reviewed-by: Thomas Huth <thuth@redhat.com>
> Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

> ---
>  Makefile.objs | 6 ++----
>  1 file changed, 2 insertions(+), 4 deletions(-)
>
> diff --git a/Makefile.objs b/Makefile.objs
> index 395dd1e670..c6321d0465 100644
> --- a/Makefile.objs
> +++ b/Makefile.objs
> @@ -71,11 +71,9 @@ qemu-seccomp.o-libs := $(SECCOMP_LIBS)
>  
>  common-obj-$(CONFIG_FDT) += device_tree.o
>  
> -######################################################################
> -# qapi
> -
>  common-obj-y += qapi/
> -endif
> +
> +endif # CONFIG_SOFTMMU
>  
>  #######################################################################
>  # Target-independent parts used in system and user emulation


-- 
Alex Bennée

[PATCH v2 5/6] hw/core: Restrict reset handlers API to system-mode

[Philippe Mathieu-Daudé]

The user-mode code does not use this API, restrict it
to the system-mode.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
---
 hw/core/Makefile.objs | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/hw/core/Makefile.objs b/hw/core/Makefile.objs
index 0edd9e635d..2fea68ccf7 100644
--- a/hw/core/Makefile.objs
+++ b/hw/core/Makefile.objs
@@ -1,6 +1,7 @@
 # core qdev-related obj files, also used by *-user:
 common-obj-y += qdev.o qdev-properties.o
-common-obj-y += bus.o reset.o
+common-obj-y += bus.o
+common-obj-$(CONFIG_SOFTMMU) += reset.o
 common-obj-$(CONFIG_SOFTMMU) += qdev-fw.o
 common-obj-$(CONFIG_SOFTMMU) += fw-path-provider.o
 # irq.o needed for qdev GPIO handling:
-- 
2.21.1

Re: [PATCH v2 5/6] hw/core: Restrict reset handlers API to system-mode

[Thomas Huth]

On 18/01/2020 15.06, Philippe Mathieu-Daudé wrote:
> The user-mode code does not use this API, restrict it
> to the system-mode.
> 
> Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
> ---
>  hw/core/Makefile.objs | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/hw/core/Makefile.objs b/hw/core/Makefile.objs
> index 0edd9e635d..2fea68ccf7 100644
> --- a/hw/core/Makefile.objs
> +++ b/hw/core/Makefile.objs
> @@ -1,6 +1,7 @@
>  # core qdev-related obj files, also used by *-user:
>  common-obj-y += qdev.o qdev-properties.o
> -common-obj-y += bus.o reset.o
> +common-obj-y += bus.o
> +common-obj-$(CONFIG_SOFTMMU) += reset.o
>  common-obj-$(CONFIG_SOFTMMU) += qdev-fw.o
>  common-obj-$(CONFIG_SOFTMMU) += fw-path-provider.o
>  # irq.o needed for qdev GPIO handling:
> 

Reviewed-by: Thomas Huth <thuth@redhat.com>

Re: [PATCH v2 5/6] hw/core: Restrict reset handlers API to system-mode

[Alex Bennée]

Philippe Mathieu-Daudé <philmd@redhat.com> writes:

> The user-mode code does not use this API, restrict it
> to the system-mode.
>
> Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
> ---
>  hw/core/Makefile.objs | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/hw/core/Makefile.objs b/hw/core/Makefile.objs
> index 0edd9e635d..2fea68ccf7 100644
> --- a/hw/core/Makefile.objs
> +++ b/hw/core/Makefile.objs
> @@ -1,6 +1,7 @@
>  # core qdev-related obj files, also used by *-user:
>  common-obj-y += qdev.o qdev-properties.o
> -common-obj-y += bus.o reset.o
> +common-obj-y += bus.o
> +common-obj-$(CONFIG_SOFTMMU) += reset.o

This seems a very minor tweaks as far as it goes. I though the only
thing needed in hw was hw/core/cpu and everything else was system
emulation?

However it at least moves the needle in the right direction:

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

>  common-obj-$(CONFIG_SOFTMMU) += qdev-fw.o
>  common-obj-$(CONFIG_SOFTMMU) += fw-path-provider.o
>  # irq.o needed for qdev GPIO handling:


-- 
Alex Bennée

Re: [PATCH v2 5/6] hw/core: Restrict reset handlers API to system-mode

[Philippe Mathieu-Daudé]

On 1/20/20 12:07 PM, Alex Bennée wrote:
> 
> Philippe Mathieu-Daudé <philmd@redhat.com> writes:
> 
>> The user-mode code does not use this API, restrict it
>> to the system-mode.
>>
>> Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
>> ---
>>   hw/core/Makefile.objs | 3 ++-
>>   1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/hw/core/Makefile.objs b/hw/core/Makefile.objs
>> index 0edd9e635d..2fea68ccf7 100644
>> --- a/hw/core/Makefile.objs
>> +++ b/hw/core/Makefile.objs
>> @@ -1,6 +1,7 @@
>>   # core qdev-related obj files, also used by *-user:
>>   common-obj-y += qdev.o qdev-properties.o
>> -common-obj-y += bus.o reset.o
>> +common-obj-y += bus.o
>> +common-obj-$(CONFIG_SOFTMMU) += reset.o
> 
> This seems a very minor tweaks as far as it goes. I though the only
> thing needed in hw was hw/core/cpu and everything else was system
> emulation?

Unfortunately qdev.o pulls in a lot of unnecessary code (qbus, machine 
properties...).

> 
> However it at least moves the needle in the right direction:
> 
> Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

Thanks!

>>   common-obj-$(CONFIG_SOFTMMU) += qdev-fw.o
>>   common-obj-$(CONFIG_SOFTMMU) += fw-path-provider.o
>>   # irq.o needed for qdev GPIO handling:
> 
> 

[PATCH v2 6/6] hw/core/Makefile: Group generic objects versus system-mode objects

[Philippe Mathieu-Daudé]

To ease review/modifications of this Makefile, group generic
objects first, then system-mode specific ones, and finally
peripherals (which are only used in system-mode).

No logical changes introduced here.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
---
 hw/core/Makefile.objs | 28 ++++++++++++++--------------
 1 file changed, 14 insertions(+), 14 deletions(-)

diff --git a/hw/core/Makefile.objs b/hw/core/Makefile.objs
index 2fea68ccf7..a522b7297d 100644
--- a/hw/core/Makefile.objs
+++ b/hw/core/Makefile.objs
@@ -1,32 +1,32 @@
 # core qdev-related obj files, also used by *-user:
 common-obj-y += qdev.o qdev-properties.o
 common-obj-y += bus.o
+common-obj-y += cpu.o
+common-obj-y += hotplug.o
+common-obj-y += vmstate-if.o
+# irq.o needed for qdev GPIO handling:
+common-obj-y += irq.o
+
 common-obj-$(CONFIG_SOFTMMU) += reset.o
 common-obj-$(CONFIG_SOFTMMU) += qdev-fw.o
 common-obj-$(CONFIG_SOFTMMU) += fw-path-provider.o
-# irq.o needed for qdev GPIO handling:
-common-obj-y += irq.o
-common-obj-y += hotplug.o
 common-obj-$(CONFIG_SOFTMMU) += nmi.o
 common-obj-$(CONFIG_SOFTMMU) += vm-change-state-handler.o
-common-obj-y += cpu.o
-common-obj-y += vmstate-if.o
+common-obj-$(CONFIG_SOFTMMU) += qdev-properties-system.o
+common-obj-$(CONFIG_SOFTMMU) += sysbus.o
+common-obj-$(CONFIG_SOFTMMU) += machine.o
+common-obj-$(CONFIG_SOFTMMU) += null-machine.o
+common-obj-$(CONFIG_SOFTMMU) += loader.o
+common-obj-$(CONFIG_SOFTMMU) += machine-hmp-cmds.o
+obj-$(CONFIG_SOFTMMU) += machine-qmp-cmds.o
+obj-$(CONFIG_SOFTMMU) += numa.o
 
 common-obj-$(CONFIG_EMPTY_SLOT) += empty_slot.o
 common-obj-$(CONFIG_XILINX_AXI) += stream.o
 common-obj-$(CONFIG_PTIMER) += ptimer.o
-common-obj-$(CONFIG_SOFTMMU) += sysbus.o
-common-obj-$(CONFIG_SOFTMMU) += machine.o
-common-obj-$(CONFIG_SOFTMMU) += loader.o
 common-obj-$(CONFIG_FITLOADER) += loader-fit.o
-common-obj-$(CONFIG_SOFTMMU) += qdev-properties-system.o
 common-obj-$(CONFIG_REGISTER) += register.o
 common-obj-$(CONFIG_OR_IRQ) += or-irq.o
 common-obj-$(CONFIG_SPLIT_IRQ) += split-irq.o
 common-obj-$(CONFIG_PLATFORM_BUS) += platform-bus.o
 common-obj-$(CONFIG_GENERIC_LOADER) += generic-loader.o
-common-obj-$(CONFIG_SOFTMMU) += null-machine.o
-
-obj-$(CONFIG_SOFTMMU) += machine-qmp-cmds.o
-obj-$(CONFIG_SOFTMMU) += numa.o
-common-obj-$(CONFIG_SOFTMMU) += machine-hmp-cmds.o
-- 
2.21.1

Re: [PATCH v2 6/6] hw/core/Makefile: Group generic objects versus system-mode objects

[Alex Bennée]

Philippe Mathieu-Daudé <philmd@redhat.com> writes:

> To ease review/modifications of this Makefile, group generic
> objects first, then system-mode specific ones, and finally
> peripherals (which are only used in system-mode).
>
> No logical changes introduced here.
>
> Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
> ---
>  hw/core/Makefile.objs | 28 ++++++++++++++--------------
>  1 file changed, 14 insertions(+), 14 deletions(-)
>
> diff --git a/hw/core/Makefile.objs b/hw/core/Makefile.objs
> index 2fea68ccf7..a522b7297d 100644
> --- a/hw/core/Makefile.objs
> +++ b/hw/core/Makefile.objs
> @@ -1,32 +1,32 @@
>  # core qdev-related obj files, also used by *-user:
>  common-obj-y += qdev.o qdev-properties.o
>  common-obj-y += bus.o
> +common-obj-y += cpu.o
> +common-obj-y += hotplug.o
> +common-obj-y += vmstate-if.o
> +# irq.o needed for qdev GPIO handling:

how are IRQs relevant to linux-user? or vmstate and hotplug?

> +common-obj-y += irq.o
> +
>  common-obj-$(CONFIG_SOFTMMU) += reset.o
>  common-obj-$(CONFIG_SOFTMMU) += qdev-fw.o
>  common-obj-$(CONFIG_SOFTMMU) += fw-path-provider.o
> -# irq.o needed for qdev GPIO handling:
> -common-obj-y += irq.o
> -common-obj-y += hotplug.o
>  common-obj-$(CONFIG_SOFTMMU) += nmi.o
>  common-obj-$(CONFIG_SOFTMMU) += vm-change-state-handler.o
> -common-obj-y += cpu.o
> -common-obj-y += vmstate-if.o
> +common-obj-$(CONFIG_SOFTMMU) += qdev-properties-system.o
> +common-obj-$(CONFIG_SOFTMMU) += sysbus.o
> +common-obj-$(CONFIG_SOFTMMU) += machine.o
> +common-obj-$(CONFIG_SOFTMMU) += null-machine.o
> +common-obj-$(CONFIG_SOFTMMU) += loader.o
> +common-obj-$(CONFIG_SOFTMMU) += machine-hmp-cmds.o
> +obj-$(CONFIG_SOFTMMU) += machine-qmp-cmds.o
> +obj-$(CONFIG_SOFTMMU) += numa.o
>  
>  common-obj-$(CONFIG_EMPTY_SLOT) += empty_slot.o
>  common-obj-$(CONFIG_XILINX_AXI) += stream.o
>  common-obj-$(CONFIG_PTIMER) += ptimer.o
> -common-obj-$(CONFIG_SOFTMMU) += sysbus.o
> -common-obj-$(CONFIG_SOFTMMU) += machine.o
> -common-obj-$(CONFIG_SOFTMMU) += loader.o
>  common-obj-$(CONFIG_FITLOADER) += loader-fit.o
> -common-obj-$(CONFIG_SOFTMMU) += qdev-properties-system.o
>  common-obj-$(CONFIG_REGISTER) += register.o
>  common-obj-$(CONFIG_OR_IRQ) += or-irq.o
>  common-obj-$(CONFIG_SPLIT_IRQ) += split-irq.o
>  common-obj-$(CONFIG_PLATFORM_BUS) += platform-bus.o
>  common-obj-$(CONFIG_GENERIC_LOADER) += generic-loader.o
> -common-obj-$(CONFIG_SOFTMMU) += null-machine.o
> -
> -obj-$(CONFIG_SOFTMMU) += machine-qmp-cmds.o
> -obj-$(CONFIG_SOFTMMU) += numa.o
> -common-obj-$(CONFIG_SOFTMMU) += machine-hmp-cmds.o


-- 
Alex Bennée

Re: [PATCH v2 0/6] buildsys: Build faster (mostly tools and linux-user)

[Philippe Mathieu-Daudé]

On 1/18/20 3:06 PM, Philippe Mathieu-Daudé wrote:
> In some configuration (linux-user, tools) we can ignore building
> various objects (and the libfdt).
> 
> Tested with all the combinations of --[enable|disable]-tools,
> --[enable|disable]-user and --[enable|disable]-system using the
> following commands (suggested by Laurent Vivier in v1):
> 
>    $ mkdir build
>    $ cd build
>    $ for user in enable disable; do \
>          for tools in enable disable; do \
>              for system in enable disable; do \
>                  rm -fr build-$user-$system-$tools && \
>                  mkdir build-$user-$system-$tools && \
>                      (cd build-$user-$system-$tools && \
>                       ../../configure \
>                                       --${user}-user \
>                                       --${system}-system \
>                                       --${tools}-tools \
>                                       --disable-docs \
>                      ); \
>              done; \
>          done; \
>      done
> 
> Then building each of the 8 subdirectories on x86_64 and aarch64
> hosts, running 'make check', and only on x86_64:
> 'make run-tcg-tests-x86_64-linux-user'.
> 
> All CI green:
> https://gitlab.com/philmd/qemu/pipelines/110420332
> https://travis-ci.org/philmd/qemu/builds/638781159
> https://app.shippable.com/github/philmd/qemu/runs/587/summary/console
> 
> Since v1:
> - no code change, improved commit description, added review tags
> - added 2 new patches touching hw/core/ (remove reset.o from linux-user)
> 
> $ git backport-diff -u v1
> Key:
> [----] : patches are identical
> [####] : number of functional differences between upstream/downstream patch
> [down] : patch is downstream-only
> The flags [FC] indicate (F)unctional and (C)ontextual differences, respectively
> 
> 001/6:[----] [--] 'configure: Do not build libfdt if not required'
> 002/6:[----] [--] 'Makefile: Clarify all the codebase requires qom/ objects'
> 003/6:[----] [--] 'Makefile: Restrict system emulation and tools objects'
> 004/6:[----] [--] 'Makefile: Remove unhelpful comment'
> 005/6:[down] 'hw/core: Restrict reset handlers API to system-mode'
> 006/6:[down] 'hw/core/Makefile: Group generic objects versus system-mode objects'
> 
> Supersedes: <20200109153939.27173-1-philmd@redhat.com>

   ^ testing latest patchew feature, v2 has different subject name
     than v1, but patchew successfully linked them :)

See "Diff against v1" in v2:
https://patchew.org/QEMU/20200118140619.26333-1-philmd@redhat.com/

https://patchew.org/QEMU/20200109153939.27173-1-philmd@redhat.com/diff/20200118140619.26333-1-philmd@redhat.com/

Thanks Kevin & Paolo for this feature :)

Re: [PATCH v2 0/6] buildsys: Build faster (mostly tools and linux-user)

[Paolo Bonzini]

On 18/01/20 15:22, Philippe Mathieu-Daudé wrote:
>>
>> Supersedes: <20200109153939.27173-1-philmd@redhat.com>
> 
>   ^ testing latest patchew feature, v2 has different subject name
>     than v1, but patchew successfully linked them :)

Appreciated!  You've now demonstrated the case where the header is
included in the cover letter rather than a separate message.  :)

Series queued, thanks.

Paolo

> See "Diff against v1" in v2:
> https://patchew.org/QEMU/20200118140619.26333-1-philmd@redhat.com/
> 
> https://patchew.org/QEMU/20200109153939.27173-1-philmd@redhat.com/diff/20200118140619.26333-1-philmd@redhat.com/
> 
> 
> Thanks Kevin & Paolo for this feature :)