Re: linux-next: build warning after merge of the ftrace tree

[Steven Rostedt]

On Wed, 15 May 2024 00:49:31 -0400
Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:

> > kernel/trace/ring_buffer.c:6286:72: warning: passing argument 1 of 'virt_to_pfn' makes pointer from integer without a cast [-Wint-conversion]
> >  6286 |                 struct page *page = virt_to_page(cpu_buffer->subbuf_ids[s]);
> >       |                                                  ~~~~~~~~~~~~~~~~~~~~~~^~~
> >       |                                                                        |
> >       |                                                                        long unsigned int
> > include/asm-generic/memory_model.h:37:45: note: in definition of macro '__pfn_to_page'
> >    37 | #define __pfn_to_page(pfn)      (vmemmap + (pfn))
> >       |                                             ^~~
> > kernel/trace/ring_buffer.c:6286:37: note: in expansion of macro 'virt_to_page'
> >  6286 |                 struct page *page = virt_to_page(cpu_buffer->subbuf_ids[s]);
> >       |                                     ^~~~~~~~~~~~
> > arch/powerpc/include/asm/page.h:228:53: note: expected 'const void *' but argument is of type 'long unsigned int'
> >   228 | static inline unsigned long virt_to_pfn(const void *kaddr)
> >       |                                         ~~~~~~~~~~~~^~~~~
> > 
> > Introduced by commit
> > 
> >   117c39200d9d ("ring-buffer: Introducing ring-buffer mapping functions")
> > 
> > My arm multi_v7_defconfig build produced a similar warning.
> > 
> > Is this really intended for v6.10?  It seems a bit late.
> >   
> 
> Well, I submitted this for the v6.9 merge window, and Linus had issues
> with it. So we've been tweaking it for the entire time and it was ready
> a bit earlier, but due to my vacation and traveling I missed pushing it
> to next. :-p
> 
> Most the code has been well tested, but because it is late, I kept it
> as a separate topic branch in case Linus still isn't happy with it.

And it looks like one of the "tweaks" dropped a (void *) cast.

This should fixed it:

diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index a02c7a52a0f5..7345a8b625fb 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -6283,7 +6283,7 @@ static int __rb_map_vma(struct ring_buffer_per_cpu *cpu_buffer,
 	}
 
 	while (p < nr_pages) {
-		struct page *page = virt_to_page(cpu_buffer->subbuf_ids[s]);
+		struct page *page = virt_to_page((void *)cpu_buffer->subbuf_ids[s]);
 		int off = 0;
 
 		if (WARN_ON_ONCE(s >= nr_subbufs)) {


-- Steve