[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH v3] target: ppc: Use MSR_HVB bit to get the target endianness

From: Narayana Murty N
Subject: Re: [PATCH v3] target: ppc: Use MSR_HVB bit to get the target endianness for memory dump
Date: Thu, 25 May 2023 09:45:11 +0530
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101 Thunderbird/102.10.0

On 5/23/23 15:52, Cédric Le Goater wrote:
On 5/22/23 18:02, Narayana Murty N wrote:
Currently on PPC64 qemu always dumps the guest memory in
Big Endian (BE) format even though the guest running in Little Endian

The patch is surely correct. I have problems understanding the config
you are testing. PPC Book3s has multiple hypervisor implementations :

1. pHyp (AKA PowerVM)
2. OPAL/PowerNV (AKA Power KVM-HV)
3. OPAL/PowerNV/pSeries (AKA Power KVMHV-on-pSeries)
4. pHyp/pSeries (very recent implementation, I don't know how it is
   referred to in the kernel)

I am leaving the KVM-PR implementation out of the discussions for

QEMU also supports emulation of 2. and 3. in two different machines
PowerNV and pseries, although running pseries guests under a PowerNV
machine is slow, so is running pseries guests under pseries.

Could you please describe your environment ?



It had been tested target machine OPAL/PowerNV with big endian host os.
and also target OPAL/PowerNV/pSeries little endian host setup with qemu. 

(LE) mode. So crash tool fails to load the dump as illustrated below:

Log :
$ virsh dump DOMAIN --memory-only dump.file

Domain 'DOMAIN' dumped to dump.file

$ crash vmlinux dump.file

crash 8.0.2-1.el9

WARNING: endian mismatch:
           crash utility: little-endian
           dump.file: big-endian

WARNING: machine type mismatch:
           crash utility: PPC64
           dump.file: (unknown)

crash: dump.file: not a supported file format

This happens because cpu_get_dump_info() passes cpu->env->has_hv_mode
to function ppc_interrupts_little_endian(), the cpu->env->has_hv_mode
always set for powerNV even though the guest is not running in hv mode.
The hv mode should be taken from msr_mask MSR_HVB bit
(cpu->env.msr_mask & MSR_HVB). This patch fixes the issue by passing
MSR_HVB value to ppc_interrupts_little_endian() in order to determine
the guest endianness.

The crash tool also expects guest kernel endianness should match the
endianness of the dump.

The patch was tested on POWER9 box booted with Linux as host in
following cases:

Host-Endianess Qemu-Target-Machine Qemu-Guest-Endianess  Qemu-Generated-Guest
BE             powernv             LE KVM guest                 LE
BE             powernv             BE KVM guest                 BE
LE             powernv             LE KVM guest                 LE
LE             powernv             BE KVM guest                 BE
LE             pseries KVM         LE KVM guest                 LE
LE             pseries TCG         LE guest                     LE

Signed-off-by: Narayana Murty N <nnmlinux@linux.ibm.com>
Changes since V2:
commit message modified as per feedbak from Nicholas Piggin.
Changes since V1:
The approach to solve the issue was changed based on feedback from
Fabiano Rosas on patch V1.
  target/ppc/arch_dump.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/target/ppc/arch_dump.c b/target/ppc/arch_dump.c
index f58e6359d5..a8315659d9 100644
--- a/target/ppc/arch_dump.c
+++ b/target/ppc/arch_dump.c
@@ -237,7 +237,7 @@ int cpu_get_dump_info(ArchDumpInfo *info,
      info->d_machine = PPC_ELF_MACHINE;
      info->d_class = ELFCLASS;
  -    if (ppc_interrupts_little_endian(cpu, cpu->env.has_hv_mode)) {
+    if (ppc_interrupts_little_endian(cpu, !!(cpu->env.msr_mask & MSR_HVB))) {
          info->d_endian = ELFDATA2LSB;
      } else {
          info->d_endian = ELFDATA2MSB;

reply via email to

[Prev in Thread] Current Thread [Next in Thread]