[kvm-unit-tests PATCH 0/2] configure: Support CONFIG_* extension

[kvm-unit-tests PATCH 0/2] configure: Support CONFIG_* extension

[Andrew Jones]

riscv already wants to change NR_CPUS sometimes, so a CONFIG_NR_CPUS,
which can be changed at compile time, would be helpful. We could just
provide support for that with the configure command line, but other
configurations will eventually come along (there are already two more
in another riscv branch). Add support for extending config.h with a
manually provided config.h-type file and also add support for using it
for CONFIG_NR_CPUS for riscv.

Andrew Jones (2):
  configure: Introduce add-config
  riscv: Make NR_CPUS configurable

 configure             | 17 ++++++++++++++++-
 lib/riscv/asm/setup.h |  3 ++-
 2 files changed, 18 insertions(+), 2 deletions(-)

-- 
2.46.0

[kvm-unit-tests PATCH 1/2] configure: Introduce add-config

[Andrew Jones]

Allow users to add additional CONFIG_* and override defaults
by concatenating a given file with #define's and #undef's to
lib/config.h

Signed-off-by: Andrew Jones <andrew.jones@linux.dev>
---
 configure | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/configure b/configure
index 27ae9cc89657..7a1317d0650d 100755
--- a/configure
+++ b/configure
@@ -64,6 +64,8 @@ usage() {
 	                           no environ is provided by the user (enabled by default)
 	    --erratatxt=FILE       specify a file to use instead of errata.txt. Use
 	                           '--erratatxt=' to ensure no file is used.
+	    --add-config=FILE      specify a file containing configs (CONFIG_*) to add on to the
+	                           generated lib/config.h. Use #undef to override default configs.
 	    --host-key-document=HOST_KEY_DOCUMENT
 	                           Specify the machine-specific host-key document for creating
 	                           a PVM image with 'genprotimg' (s390x only)
@@ -153,6 +155,10 @@ while [[ "$1" = -* ]]; do
 	    erratatxt=
 	    [ "$arg" ] && erratatxt=$(eval realpath "$arg")
 	    ;;
+	--add-config)
+	    add_config=
+	    [ "$arg" ] && add_config=$(eval realpath "$arg")
+	    ;;
 	--host-key-document)
 	    host_key_document="$arg"
 	    ;;
@@ -213,6 +219,10 @@ if [ "$erratatxt" ] && [ ! -f "$erratatxt" ]; then
     echo "erratatxt: $erratatxt does not exist or is not a regular file"
     exit 1
 fi
+if [ "$add_config" ] && [ ! -f "$add_config" ]; then
+    echo "add-config: $add_config does not exist or is not a regular file"
+    exit 1
+fi
 
 arch_name=$arch
 [ "$arch" = "aarch64" ] && arch="arm64"
@@ -502,4 +512,8 @@ cat <<EOF >> lib/config.h
 
 EOF
 fi
+if [ "$add_config" ]; then
+    echo "/* Additional configs from $add_config */" >> lib/config.h
+    cat "$add_config" >> lib/config.h
+fi
 echo "#endif" >> lib/config.h
-- 
2.46.0

[kvm-unit-tests PATCH 2/2] riscv: Make NR_CPUS configurable

[Andrew Jones]

Unit tests would like to go nuts with the number of harts in order
to help shake out issues with hart number assumptions. Rather than
set a huge number that will only be used when a platform supports
a huge number or when QEMU is told to exceed the recommended
number of vcpus, make the number configurable. However, we do bump
the default from 16 to 2*xlen since we would like to always force
kvm-unit-tests to use cpumasks with more than one word in order to
ensure that code stays maintained.

To override the default for NR_CPUS to, e.g. 256, testers should use
--add-config. For example,

  $ cat <<EOF > 256.config
  #undef CONFIG_NR_CPUS
  #define CONFIG_NR_CPUS 256
  EOF
  $ ./configure --arch=riscv64 --cross-prefix=riscv64-linux-gnu- --add-config=256.config

Signed-off-by: Andrew Jones <andrew.jones@linux.dev>
---
 configure             | 3 ++-
 lib/riscv/asm/setup.h | 3 ++-
 2 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/configure b/configure
index 7a1317d0650d..5ed0c28fcaea 100755
--- a/configure
+++ b/configure
@@ -508,7 +508,8 @@ EOF
 elif [ "$arch" = "riscv32" ] || [ "$arch" = "riscv64" ]; then
 cat <<EOF >> lib/config.h
 
-#define CONFIG_UART_EARLY_BASE 0x10000000
+#define CONFIG_NR_CPUS			(__riscv_xlen * 2)
+#define CONFIG_UART_EARLY_BASE		0x10000000
 
 EOF
 fi
diff --git a/lib/riscv/asm/setup.h b/lib/riscv/asm/setup.h
index a13159bfe395..43b63c56d96f 100644
--- a/lib/riscv/asm/setup.h
+++ b/lib/riscv/asm/setup.h
@@ -2,9 +2,10 @@
 #ifndef _ASMRISCV_SETUP_H_
 #define _ASMRISCV_SETUP_H_
 #include <libcflat.h>
+#include <config.h>
 #include <asm/processor.h>
 
-#define NR_CPUS 16
+#define NR_CPUS CONFIG_NR_CPUS
 extern struct thread_info cpus[NR_CPUS];
 extern int nr_cpus;
 extern uint64_t timebase_frequency;
-- 
2.46.0

Re: [kvm-unit-tests PATCH 1/2] configure: Introduce add-config

[Nicholas Piggin]

On Wed Sep 4, 2024 at 12:39 AM AEST, Andrew Jones wrote:
> Allow users to add additional CONFIG_* and override defaults
> by concatenating a given file with #define's and #undef's to
> lib/config.h

That's a horrible config format lol, but probbaly the simplest way to
get something working. What if you included the user config first, then
make the generated config test ifndef before defining the default?

Is it better to have a config file than to just add more --options to
configure? If we had thousands of options maybe, but so far we are
getting by with configure options. I think I prefer that for now
unless we wholesale moved everything to a .config style.

Thanks,
Nick

>
> Signed-off-by: Andrew Jones <andrew.jones@linux.dev>
> ---
>  configure | 14 ++++++++++++++
>  1 file changed, 14 insertions(+)
>
> diff --git a/configure b/configure
> index 27ae9cc89657..7a1317d0650d 100755
> --- a/configure
> +++ b/configure
> @@ -64,6 +64,8 @@ usage() {
>  	                           no environ is provided by the user (enabled by default)
>  	    --erratatxt=FILE       specify a file to use instead of errata.txt. Use
>  	                           '--erratatxt=' to ensure no file is used.
> +	    --add-config=FILE      specify a file containing configs (CONFIG_*) to add on to the
> +	                           generated lib/config.h. Use #undef to override default configs.
>  	    --host-key-document=HOST_KEY_DOCUMENT
>  	                           Specify the machine-specific host-key document for creating
>  	                           a PVM image with 'genprotimg' (s390x only)
> @@ -153,6 +155,10 @@ while [[ "$1" = -* ]]; do
>  	    erratatxt=
>  	    [ "$arg" ] && erratatxt=$(eval realpath "$arg")
>  	    ;;
> +	--add-config)
> +	    add_config=
> +	    [ "$arg" ] && add_config=$(eval realpath "$arg")
> +	    ;;
>  	--host-key-document)
>  	    host_key_document="$arg"
>  	    ;;
> @@ -213,6 +219,10 @@ if [ "$erratatxt" ] && [ ! -f "$erratatxt" ]; then
>      echo "erratatxt: $erratatxt does not exist or is not a regular file"
>      exit 1
>  fi
> +if [ "$add_config" ] && [ ! -f "$add_config" ]; then
> +    echo "add-config: $add_config does not exist or is not a regular file"
> +    exit 1
> +fi
>  
>  arch_name=$arch
>  [ "$arch" = "aarch64" ] && arch="arm64"
> @@ -502,4 +512,8 @@ cat <<EOF >> lib/config.h
>  
>  EOF
>  fi
> +if [ "$add_config" ]; then
> +    echo "/* Additional configs from $add_config */" >> lib/config.h
> +    cat "$add_config" >> lib/config.h
> +fi
>  echo "#endif" >> lib/config.h

Re: [kvm-unit-tests PATCH 1/2] configure: Introduce add-config

[Andrew Jones]

On Wed, Sep 11, 2024 at 10:39:03AM GMT, Nicholas Piggin wrote:
> On Wed Sep 4, 2024 at 12:39 AM AEST, Andrew Jones wrote:
> > Allow users to add additional CONFIG_* and override defaults
> > by concatenating a given file with #define's and #undef's to
> > lib/config.h
> 
> That's a horrible config format lol, but probbaly the simplest way to
> get something working. What if you included the user config first, then
> make the generated config test ifndef before defining the default?

User config first and then #ifndef would indeed be better.

> 
> Is it better to have a config file than to just add more --options to
> configure? If we had thousands of options maybe, but so far we are
> getting by with configure options.

I have some unposted patches where I introduce two more configs, which
is what inspired me to stop adding configure command line options.

> I think I prefer that for now
> unless we wholesale moved everything to a .config style.

Moving to .config would be good, and importing and applying Kconfiglib
doesn't look too daunting either. We can put this --add-config idea on
hold until we've had a chance to experiment.

Thanks,
drew