From mboxrd@z Thu Jan 1 00:00:00 1970 From: hanjun.guo@linaro.org (Hanjun Guo) Date: Thu, 28 May 2015 11:22:57 +0800 Subject: ARM64 kexec/kdump timeline In-Reply-To: References: <555E37FC.9020800@codeaurora.org> <1432239334.1922.7.camel@infradead.org> <55653538.8020206@codeaurora.org> Message-ID: <55668A11.1010304@linaro.org> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On 2015?05?27? 17:38, Ard Biesheuvel wrote: > On 27 May 2015 at 05:08, Timur Tabi wrote: >> On 05/21/2015 03:15 PM, Geoff Levand wrote: >>> >>> I feel kexec was ready to merge nine months ago. Users have only sent >>> private e-mails like yourself, so there has been no public inquiries, >>> and so, the arm64 maintainers think kexec is not of interest to users. >> >> >> Does kexec on ARM64 support ACPI? In some of our initial tests, it appears >> that the kexec'd kernel won't boot because it's expecting a device tree, and >> we don't have one. >> >> ARM64 servers are supposed to use ACPI instead of a device tree. >> > > The latest arm64 kexec patches support booting via UEFI, and when > booting via UEFI, the ACPI root pointer is retrieved from a UEFI > configuration table if the FDT contains no system description. So ACPI > support should not depend at all on how kexec is implemented (unless > your FDT does contain a system description, in which case you should > pass along the 'acpi=xxx' param that was given to the cold booted > kernel) Thanks for the clarify. Geoff also tested kexec on FVP with ACPI enabled and it works as far as I know. Timur, can you provide more detailed information then we can debug into it? Thanks Hanjun