[RFC PATCH] powerpc/kexec: Add kexec "hold" support for Book3e processors
jimix at pobox.com
Tue Dec 4 14:05:47 EST 2012
IBM Blue Gene/Q comes with some very strange firmware that I'm trying to get out of using in the kernel. So instead I spin all the threads in the boot wrapper (using the firmware) and have them enter the kexec stub, pre-translated at the virtual "linear" address, never touching firmware again.
This works strategy works wonderfully, but I need the following patch in the kexec stub.
I believe it should not effect Book3S and Book3E does not appear to be here yet so I'd love to get any criticisms up front.
This patch adds two items:
1) Book3e requires that GPR4 survive the "hold" process, so we make
sure that happens.
2) Book3e has no real mode, and the hold code exploits this. Since
these processors ares always translated, we arrange for the kexeced
threads to enter the hold code using the normal kernel linear mapping.
Signed-off-by: Jimi Xenidis <jimix at pobox.com>
arch/powerpc/kernel/head_64.S | 19 ++++++++++++++++++-
1 files changed, 18 insertions(+), 1 deletions(-)
diff --git a/arch/powerpc/kernel/head_64.S b/arch/powerpc/kernel/head_64.S
index 58bddee..1b00e95 100644
@@ -122,6 +122,8 @@ __secondary_hold:
/* Grab our physical cpu number */
+ /* stash r4 for book3e */
+ mr r25,r4
/* Tell the master cpu we're here */
/* Relocation is off & we are located at an address less */
@@ -129,16 +131,31 @@ __secondary_hold:
+ li r26,0
/* All secondary cpus wait here until told to start. */
-100: ld r4,__secondary_hold_spinloop-_stext(0)
+100: ld r4,__secondary_hold_spinloop-_stext(r26)
#if defined(CONFIG_SMP) || defined(CONFIG_KEXEC)
ld r4,0(r4) /* deref function descriptor */
+ * it may be the case that other platforms have r4 right to
+ * begin with, this gives us some safety in case it is not
+ mr r4,r25
/* Make sure that patched code is visible */
More information about the Linuxppc-dev