* Re: [PATCH] KVM/MIPS32: Export min_low_pfn.
2013-05-17 21:42 ` David Daney
@ 2013-05-18 6:36 ` Gleb Natapov
2013-05-19 9:22 ` Ralf Baechle
1 sibling, 0 replies; 4+ messages in thread
From: Gleb Natapov @ 2013-05-18 6:36 UTC (permalink / raw)
To: David Daney; +Cc: Sanjay Lal, linux-mips, kvm, ralf, mtosatti
On Fri, May 17, 2013 at 02:42:48PM -0700, David Daney wrote:
> On 05/17/2013 02:06 PM, Sanjay Lal wrote:
> >The KVM module uses the standard MIPS cache management routines, which use min_low_pfn.
> >This creates and indirect dependency, requiring min_low_pfn to be exported.
> >
> >Signed-off-by: Sanjay Lal <sanjayl@kymasys.com>
> >---
> > arch/mips/kernel/mips_ksyms.c | 6 ++++++
> > 1 file changed, 6 insertions(+)
> >
> >diff --git a/arch/mips/kernel/mips_ksyms.c b/arch/mips/kernel/mips_ksyms.c
> >index 6e58e97..0299472 100644
> >--- a/arch/mips/kernel/mips_ksyms.c
> >+++ b/arch/mips/kernel/mips_ksyms.c
> >@@ -14,6 +14,7 @@
> > #include <linux/mm.h>
> > #include <asm/uaccess.h>
> > #include <asm/ftrace.h>
> >+#include <linux/bootmem.h>
> >
> > extern void *__bzero(void *__s, size_t __count);
> > extern long __strncpy_from_user_nocheck_asm(char *__to,
> >@@ -60,3 +61,8 @@ EXPORT_SYMBOL(invalid_pte_table);
> > /* _mcount is defined in arch/mips/kernel/mcount.S */
> > EXPORT_SYMBOL(_mcount);
> > #endif
> >+
> >+/* The KVM module uses the standard MIPS cache functions which use
> >+ * min_low_pfn, requiring it to be exported.
> >+ */
> >+EXPORT_SYMBOL(min_low_pfn);
>
> I think I asked this before, but I don't remember the answer:
>
> Why not put EXPORT_SYMBOL(min_low_pfn) in mm/bootmem.c adjacent to
> where the symbol is defined?
>
He did answered here:
http://permalink.gmane.org/gmane.comp.emulators.kvm.devel/109895.
I suggested mips_ksyms.c solution as an option.
> Cluttering up the kernel with multiple architectures all doing
> architecture specific exports of the same symbol is not a clean way
> of doing things.
>
> The second time something needs to be done, it should be factored
> out into common code.
>
Exports are different. You define interface between the kernel and modules
here, exporting the symbol may not be desirable for some arch. And
min_low_pfn is not the only example. Anything in arch _ksyms files is like that:
exported by some archs but not the others.
--
Gleb.
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] KVM/MIPS32: Export min_low_pfn.
2013-05-17 21:42 ` David Daney
2013-05-18 6:36 ` Gleb Natapov
@ 2013-05-19 9:22 ` Ralf Baechle
1 sibling, 0 replies; 4+ messages in thread
From: Ralf Baechle @ 2013-05-19 9:22 UTC (permalink / raw)
To: David Daney, David Howells, Tony Luck, Fenghua Yu, linux-ia64,
James Hogan, Michal Simek, microblaze-uclinux, Chen Liqin,
Paul Mundt, linux-sh, linux-mm, Arnd Bergmann, linux-arch
Cc: Sanjay Lal, linux-mips, kvm, gleb, mtosatti
On Fri, May 17, 2013 at 02:42:48PM -0700, David Daney wrote:
> On 05/17/2013 02:06 PM, Sanjay Lal wrote:
> >The KVM module uses the standard MIPS cache management routines, which use min_low_pfn.
> >This creates and indirect dependency, requiring min_low_pfn to be exported.
> >
> >Signed-off-by: Sanjay Lal <sanjayl@kymasys.com>
> >---
> > arch/mips/kernel/mips_ksyms.c | 6 ++++++
> > 1 file changed, 6 insertions(+)
> >
> >diff --git a/arch/mips/kernel/mips_ksyms.c b/arch/mips/kernel/mips_ksyms.c
> >index 6e58e97..0299472 100644
> >--- a/arch/mips/kernel/mips_ksyms.c
> >+++ b/arch/mips/kernel/mips_ksyms.c
> >@@ -14,6 +14,7 @@
> > #include <linux/mm.h>
> > #include <asm/uaccess.h>
> > #include <asm/ftrace.h>
> >+#include <linux/bootmem.h>
> >
> > extern void *__bzero(void *__s, size_t __count);
> > extern long __strncpy_from_user_nocheck_asm(char *__to,
> >@@ -60,3 +61,8 @@ EXPORT_SYMBOL(invalid_pte_table);
> > /* _mcount is defined in arch/mips/kernel/mcount.S */
> > EXPORT_SYMBOL(_mcount);
> > #endif
> >+
> >+/* The KVM module uses the standard MIPS cache functions which use
> >+ * min_low_pfn, requiring it to be exported.
The comment is wrong. min_low_pfn is being referenced by pfn_valid() which
is implemented (simplified for purposes of this discussion) like:
int pfn_valid(unsigned long pfn)
{
return pfn >= min_low_pfn && pfn < max_mapnr;
}
> >+ */
> >+EXPORT_SYMBOL(min_low_pfn);
>
> I think I asked this before, but I don't remember the answer:
>
> Why not put EXPORT_SYMBOL(min_low_pfn) in mm/bootmem.c adjacent to
> where the symbol is defined?
>
> Cluttering up the kernel with multiple architectures all doing
> architecture specific exports of the same symbol is not a clean way
> of doing things.
>
> The second time something needs to be done, it should be factored
> out into common code.
pfn_valid() should return 1 if for the pfn passed as the argment a struct
page exists in mem_map[] - this only affects the flatmem case. However
it is possible that min_low_pfn is bigger than ARCH_PFN_OFFSET in which
case pfn_valid might return 0 even though it should have returned 1.
So I fixed the issue like:
static inline int pfn_valid(unsigned long pfn)
{
return pfn >= ARCH_PFN_OFFSET && pfn < max_mapnr;
}
frv, ia64, metag, microblaze, score and sh are also implementing pfn_valid
referencing min_low_pfn.
Time to move pfn_valid() to asm-generic, fix the min_low_pfn /
ARCH_PFN_OFFSET issue and while at it, harden it against multiple
evaluation of its arguments?
Ralf
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 4+ messages in thread