qemu-ppc
[Top][All Lists]
Advanced

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

Re: [PULL 02/41] target/ppc: 603: fix restore of GPRs 0-3 on rfi


From: Mark Cave-Ayland
Subject: Re: [PULL 02/41] target/ppc: 603: fix restore of GPRs 0-3 on rfi
Date: Mon, 31 Jan 2022 12:01:49 +0000
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101 Thunderbird/91.5.0

On 31/01/2022 11:07, Cédric Le Goater wrote:

From: Christophe Leroy <christophe.leroy@csgroup.eu>

After a TLB miss exception, GPRs 0-3 must be restored on rfi.

This is managed by hreg_store_msr() which is called by do_rfi()

However, hreg_store_msr() does it if MSR[TGPR] is unset in the
passed MSR value.

The problem is that do_rfi() is given the content of SRR1 as
the value to be set in MSR, but TGPR bit is not part of SRR1
and that bit is used for something else and is sometimes set
to 1, leading to hreg_store_msr() not restoring GPRs.

So, do the same way as for POW bit, force clearing it.

Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Cc: Cedric Le Goater <clg@kaod.org>
Cc: Fabiano Rosas <farosas@linux.ibm.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20220120103824.239573-1-christophe.leroy@csgroup.eu>
Signed-off-by: Cédric Le Goater <clg@kaod.org>
---
  target/ppc/excp_helper.c | 4 ++++
  1 file changed, 4 insertions(+)

diff --git a/target/ppc/excp_helper.c b/target/ppc/excp_helper.c
index bc646c67a0f5..980f62fd7964 100644
--- a/target/ppc/excp_helper.c
+++ b/target/ppc/excp_helper.c
@@ -1164,6 +1164,10 @@ static void do_rfi(CPUPPCState *env, target_ulong nip, 
target_ulong msr)
      /* MSR:POW cannot be set by any form of rfi */
      msr &= ~(1ULL << MSR_POW);
+ /* MSR:TGPR cannot be set by any form of rfi */
+    if (env->flags & POWERPC_FLAG_TGPR)
+        msr &= ~(1ULL << MSR_TGPR);
+
  #if defined(TARGET_PPC64)
      /* Switching to 32-bit ? Crop the nip */
      if (!msr_is_64bit(env, msr)) {

Have you tried a pre-PR push to Gitlab CI for your pull-ppc-20220130 tag? I'd expect this to fail the check-patch job due to the missing braces around the if() statement.


ATB,

Mark.



reply via email to

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