[v4.4,20/45] mm: Introduce lm_alias

Commit Message

Viresh Kumar June 14, 2019, 3:08 a.m. UTC

From: Laura Abbott <[e-mail]>

commit 568c5fe5a54f2654f5a4c599c45b8a62ed9a2013 upstream.

Certain architectures may have the kernel image mapped separately to
alias the linear map. Introduce a macro lm_alias to translate a kernel
image symbol into its linear alias. This is used in part with work to
add CONFIG_DEBUG_VIRTUAL support for arm64.

Reviewed-by: Mark Rutland <[e-mail]>
Tested-by: Mark Rutland <[e-mail]>
Signed-off-by: Laura Abbott <[e-mail]>
Signed-off-by: Will Deacon <[e-mail]>
Signed-off-by: Viresh Kumar <[e-mail]>
---
 include/linux/mm.h | 4 ++++
 1 file changed, 4 insertions(+)

Comments

Julien Thierry June 17, 2019, 12:33 p.m. UTC | #1

Hi Viresh,

On 14/06/2019 04:08, Viresh Kumar wrote:
> From: Laura Abbott <[e-mail]>
> 
> commit 568c5fe5a54f2654f5a4c599c45b8a62ed9a2013 upstream.
> 
> Certain architectures may have the kernel image mapped separately to
> alias the linear map. Introduce a macro lm_alias to translate a kernel
> image symbol into its linear alias. This is used in part with work to
> add CONFIG_DEBUG_VIRTUAL support for arm64.
> 

I think this commit was backported in 4.9 because one of the commits you
dropped (6840bdd73d07 arm64: KVM: Use per-CPU vector when BP hardening
is enabled) depended on it. I have yet to check whether that other
commit can be just dropped, however on your branch 4.4 branch, lm_alias
isn't used anywhere, so we probably don't want to backport this
particular patch (unless we need to actually backport the other patch in
some way).

Cheers,

Viresh Kumar June 18, 2019, 5 a.m. UTC | #2

On 17-06-19, 13:33, Julien Thierry wrote:
> I think this commit was backported in 4.9 because one of the commits you
> dropped (6840bdd73d07 arm64: KVM: Use per-CPU vector when BP hardening
> is enabled) depended on it.

Looks like that. I dropped 6840bdd73d07 patch at a later point of time when the
conflicts couldn't get resolved properly any further. I should have dropped
$subject one as well :(

> I have yet to check whether that other
> commit can be just dropped, however on your branch 4.4 branch, lm_alias
> isn't used anywhere, so we probably don't want to backport this
> particular patch (unless we need to actually backport the other patch in
> some way).

Right, dropping this doesn't generate any compilation warnings so we should be
good without it.

Patch

diff --git a/include/linux/mm.h b/include/linux/mm.h
index 251adf4d8a71..f86fdf015c74 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -59,6 +59,10 @@  extern int sysctl_legacy_va_layout;
 #define __pa_symbol(x)  __pa(RELOC_HIDE((unsigned long)(x), 0))
 #endif
 
+#ifndef lm_alias
+#define lm_alias(x)	__va(__pa_symbol(x))
+#endif
+
 /*
  * To prevent common memory management code establishing
  * a zero page mapping on a read fault.