From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from sullivan.realtime.net (sullivan.realtime.net [205.238.132.226]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client did not present a certificate) by ozlabs.org (Postfix) with ESMTP id 23E40DDF0C for ; Wed, 11 Apr 2007 18:29:16 +1000 (EST) Date: Wed, 11 Apr 2007 03:28:38 -0500 (CDT) Subject: [PATCH 0/3+2+8] powerpc zImage kexec From: Milton Miller Sender: Milton Miller To: , Message-Id: Cc: Paul Mackerras List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , The patches following this message implement support for creating a compressed zImage that can be used where a vmlinux image with a flat device-tree would be. It has been tested with kexec-tools on 64 bit platforms with upto 8 threads, and with an external bootloader. that supplies a device tree and initrd. The patches are: (1) a series of 3 patches for kexec-tools (a) a hack to fix memory corruption in the testing tree (b) a fix for @ha relocations (c) a major change to the handling of SMP slaves during purgatory (2) a related patch for increasing the times for slaves to transiton (3) a convinence patch to expose a retained initrd for testing kexec-reboot (4) and a series of 8 patches for the zImage wrapper (a) a debug message (b) smp handling code (c) coarse memory allocation for finding malloc pool, vmlinux, etc. (d) 64 bit support (e) kexec entrypoint base support -- or any environment that can supply a flat tree that fully describes the machine, and wishes to use it to find memory for the machine. (f) support for converting older versions of the flat device tree (g) support for writing to an rtas put-term-char console (h) alternative example code for using a-f on another palatform. The kernel series is based on the patches I extracted from the for-2.6.22 branch of powerpc.git as well as 1, 3, and 4 from my last series. I did not include my external payload patch at this point because I still have to debug the initramfs search code, but the plan is to allow the base code to find the vmlinux in a compressed cpio.gz in the initrd area, with the file to decompress being selected by a property in the device-tree. Because of the vaired nature of the patches, each has its own CC list. I ask that the non-bootwrapper patchs and the debug message be queued now. I expect the last one be discussion only, and the others to provoke discussion. I have been running a loop on 4 and 8 thread pSeries boxes of continous kexecs for several thousand interations now. Thanks, milton kexec-tools: ppc64-miscount-exlude-hack ppc64-reloc-ha ppc64-steal-kernel-slaves misc: initrd-debugfs kexec-slaves-earlier Patch: [PATCH 1/4] bootwrapper: missing relocation in crt0.S http://patchwork.ozlabs.org/linuxppc/patch?id=10219 Message ID Patch: Re: Patch: [PATCH 3/4] bootwrapper: no gzip fixes http://patchwork.ozlabs.org/linuxppc/patch?id=10223 Message ID Patch: Re: [PATCH 4/4] bootwrapper: decompress less, check more http://patchwork.ozlabs.org/linuxppc/patch?id=10232 Message ID boot-exact-debug boot-marshal-smp boot-memranges boot-64 boot-kexec boot-flat-conv-2-16 boot-rtas boot-bml.patch Patch: [RFC] bootwrapper: allow vmlinuz to be an external payload Submitter Milton Miller Date 2007-03-28 18:21:03 Message ID