On Fri, Dec 6, 2024 at 8:35 AM Dennis Zhou [off-list ref] wrote:

Hi Uros,

On Thu, Dec 05, 2024 at 04:40:55PM +0100, Uros Bizjak wrote:

quoted

The patch introduces per_cpu_qual define and repurposes __percpu
tag as a named address space qualifier using the new define.

Arches can now conditionally define __per_cpu_qual as their
named address space qualifier for percpu variables.

Signed-off-by: Uros Bizjak <redacted>
Acked-by: Nadav Amit <redacted>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Thomas Gleixner <redacted>
Cc: Dennis Zhou <dennis@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Christoph Lameter <redacted>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Brian Gerst <redacted>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Peter Zijlstra <peterz@infradead.org>
---
 include/asm-generic/percpu.h   | 15 +++++++++++++++
 include/linux/compiler_types.h |  2 +-
 2 files changed, 16 insertions(+), 1 deletion(-)

diff --git a/include/asm-generic/percpu.h b/include/asm-generic/percpu.h
index 50597b975a49..3b93b168faa1 100644
--- a/include/asm-generic/percpu.h
+++ b/include/asm-generic/percpu.h

@@ -6,6 +6,21 @@
 #include <linux/threads.h>
 #include <linux/percpu-defs.h>

+/*
+ * per_cpu_qual is the qualifier for the percpu named address space.
+ *
+ * Most arches use generic named address space for percpu variables but
+ * some arches define percpu variables in different named address space
+ * (on the x86 arch, percpu variable may be declared as being relative
+ * to the %fs or %gs segments using __seg_fs or __seg_gs named address
+ * space qualifier).
+ */
+#ifdef __per_cpu_qual

I read through the series and I think my only nit would be here. Can we
name this __percpu_qual? My thoughts are that it keeps it consistent
with the old address space identifier and largely most of the core
percpu stuff is defined with percpu as the naming scheme.

I based the approach on the definition of per_cpu_offset() a few lines
bellow in include/asm-generic/percpu.h:

--q--
#ifndef __per_cpu_offset
extern unsigned long __per_cpu_offset[NR_CPUS];

#define per_cpu_offset(x) (__per_cpu_offset[x])
#endif
--/q--

Sure, we can call this __percpu_qual. So, the definition in
arch/x86/include/asm/percpu.h would read as:

# define __percpu_qual      __percpu_seg_override

quoted

+# define per_cpu_qual __per_cpu_qual
+#else
+# define per_cpu_qual
+#endif

The above part could be recoded as:

#ifndef __percpu_qual
# define __percpu_qual
#endif

while the line below would become:

# define __percpu    __percpu_qual BTF_TYPE_TAG(percpu)

quoted

+
 #ifdef CONFIG_SMP

 /*

diff --git a/include/linux/compiler_types.h b/include/linux/compiler_types.h
index 981cc3d7e3aa..877fe0c43c5d 100644
--- a/include/linux/compiler_types.h
+++ b/include/linux/compiler_types.h

@@ -57,7 +57,7 @@ static inline void __chk_io_ptr(const volatile void __iomem *ptr) { }
 #  define __user     BTF_TYPE_TAG(user)
 # endif
 # define __iomem
-# define __percpu    BTF_TYPE_TAG(percpu)
+# define __percpu    per_cpu_qual BTF_TYPE_TAG(percpu)
 # define __rcu               BTF_TYPE_TAG(rcu)

 # define __chk_user_ptr(x)   (void)0

Let me test these changes a bit, I'll send a v3 in a couple of days.

Thanks,
Uros.