* [kexec-lite PATCH] trampoline: Reset primary cpu endian to big-endian
@ 2015-07-08 1:25 Samuel Mendoza-Jonas
2015-07-08 1:56 ` Anton Blanchard
0 siblings, 1 reply; 3+ messages in thread
From: Samuel Mendoza-Jonas @ 2015-07-08 1:25 UTC (permalink / raw)
To: anton; +Cc: Samuel Mendoza-Jonas, linuxppc-dev
Older big-endian ppc64 kernels don't include the FIXUP_ENDIAN check,
meaning if we kexec from a little-endian kernel the target kernel will
fail to boot.
Returning to big-endian before we enter the target kernel ensures that
the target kernel can boot whether or not it includes FIXUP_ENDIAN.
Signed-off-by: Samuel Mendoza-Jonas <sam.mj@au1.ibm.com>
---
kexec_trampoline.S | 9 ++++++++-
kexec_trampoline.h | 17 +++++++++++++++--
2 files changed, 23 insertions(+), 3 deletions(-)
diff --git a/kexec_trampoline.S b/kexec_trampoline.S
index a3eb314..29537ab 100644
--- a/kexec_trampoline.S
+++ b/kexec_trampoline.S
@@ -41,6 +41,8 @@
#define r7 7
#define r8 8
#define r9 9
+#define r10 10
+#define r11 11
#ifdef __powerpc64__
#define LOAD ld
@@ -50,8 +52,10 @@
#if defined(__LITTLE_ENDIAN__)
#define STWX_BE stwbrx
+#define BCTR .long 0x2004804e
#elif defined(__BIG_ENDIAN__)
#define STWX_BE stwx
+#define BCTR bctr
#else
#error no endianness defined!
#endif
@@ -89,7 +93,10 @@ start:
li r5,0
mtctr r4
- bctr
+
+ RESET_ENDIAN
+
+ BCTR
. = KERNEL_ADDR_OFFSET
kernel_addr:
diff --git a/kexec_trampoline.h b/kexec_trampoline.h
index 6485c88..909953b 100644
--- a/kexec_trampoline.h
+++ b/kexec_trampoline.h
@@ -27,7 +27,20 @@
/* Fixed offset in device tree for storing the physical ID of the boot CPU */
#define DT_CPU_OFFSET 28
-#ifndef __ASSEMBLY__
+#ifdef __ASSEMBLY__
+#define RESET_ENDIAN \
+ mfmsr r9; \
+ clrldi. r10,r9,63; \
+ beq $+36; /* Already big-endian */ \
+ bcl 20,31,$+4; \
+ mflr r10; \
+ addi r10,r10,28; \
+ mfmsr r11; \
+ xori r11,r11,1; \
+ mtsrr0 r10; \
+ mtsrr1 r11; \
+ rfid;
+#else
extern char __trampoline_start[];
extern char __trampoline_end[];
@@ -48,6 +61,6 @@ static inline void trampoline_set_device_tree(void *p, unsigned long addr)
*v = addr;
}
-#endif
+#endif /* !__ASSEMBLY__ */
#endif
--
2.4.5
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [kexec-lite PATCH] trampoline: Reset primary cpu endian to big-endian
2015-07-08 1:25 [kexec-lite PATCH] trampoline: Reset primary cpu endian to big-endian Samuel Mendoza-Jonas
@ 2015-07-08 1:56 ` Anton Blanchard
2015-07-08 3:11 ` Samuel Mendoza-Jonas
0 siblings, 1 reply; 3+ messages in thread
From: Anton Blanchard @ 2015-07-08 1:56 UTC (permalink / raw)
To: Samuel Mendoza-Jonas; +Cc: linuxppc-dev
Hi Sam,
> Older big-endian ppc64 kernels don't include the FIXUP_ENDIAN check,
> meaning if we kexec from a little-endian kernel the target kernel will
> fail to boot.
> Returning to big-endian before we enter the target kernel ensures that
> the target kernel can boot whether or not it includes FIXUP_ENDIAN.
Thanks!
I wonder if we can do something simpler, and always enter via rfid.
Avoids the need for the endian trampoline. Something like:
mtsrr0 r4
mfmsr r5
clrrdi r5,r5,1 /* Clear MSR_LE */
mtsrr1 r5
li r5,0
rfid
Anton
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [kexec-lite PATCH] trampoline: Reset primary cpu endian to big-endian
2015-07-08 1:56 ` Anton Blanchard
@ 2015-07-08 3:11 ` Samuel Mendoza-Jonas
0 siblings, 0 replies; 3+ messages in thread
From: Samuel Mendoza-Jonas @ 2015-07-08 3:11 UTC (permalink / raw)
To: Anton Blanchard; +Cc: linuxppc-dev
On 08/07/15 11:56, Anton Blanchard wrote:
> Hi Sam,
>
>> Older big-endian ppc64 kernels don't include the FIXUP_ENDIAN check,
>> meaning if we kexec from a little-endian kernel the target kernel will
>> fail to boot.
>> Returning to big-endian before we enter the target kernel ensures that
>> the target kernel can boot whether or not it includes FIXUP_ENDIAN.
>
> Thanks!
>
> I wonder if we can do something simpler, and always enter via rfid.
> Avoids the need for the endian trampoline. Something like:
>
> mtsrr0 r4
>
> mfmsr r5
> clrrdi r5,r5,1 /* Clear MSR_LE */
> mtsrr1 r5
>
> li r5,0
>
> rfid
>
> Anton
>
Ah that's a neat idea, I'll test and send a V2.
Cheers,
Sam
--
-----------
LTC Ozlabs
IBM
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2015-07-08 3:12 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2015-07-08 1:25 [kexec-lite PATCH] trampoline: Reset primary cpu endian to big-endian Samuel Mendoza-Jonas
2015-07-08 1:56 ` Anton Blanchard
2015-07-08 3:11 ` Samuel Mendoza-Jonas
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).