kvm.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Benjamin Herrenschmidt <benh@ozlabs.org>
To: kvm-ppc@vger.kernel.org, kvm@vger.kernel.org
Cc: Alexander Graf <agraf@suse.de>
Subject: [PATCH] powerpc/kvm: Allow qemu hypercalls to set all regs
Date: Fri, 27 Jul 2012 16:18:24 +1000	[thread overview]
Message-ID: <1343369904.2118.20.camel@pasglop> (raw)

Right now, whenever we exit for an hcall, upon return, we fetch
some register values from the structure that carries the hcall
results and update the vcpu accordingly.

However, if the hypercall chooses instead to update the registers
itself by calling set_regs, then we end up clobbering those values.

This is for example the case of the rtas calls used to reboot the
machine where a new CPU state is established and must not be clobbered.

The simple fix is to always clear the hcall_needed flag when a
set-regs happen. This fixes reboot problems.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
---

Note: There are similar issues if the reboot is triggered by an MMIO
emulation, or an OSI call, Alex, you might want to handle those cases
as well. In any cases, I'd like this to still go into 3.6 but I can
send it myself to Linus if you want.

diff --git a/arch/powerpc/kvm/book3s.c b/arch/powerpc/kvm/book3s.c
index 3f2a836..9ab13d6 100644
--- a/arch/powerpc/kvm/book3s.c
+++ b/arch/powerpc/kvm/book3s.c
@@ -463,6 +463,14 @@ int kvm_arch_vcpu_ioctl_set_regs(struct kvm_vcpu *vcpu, struct kvm_regs *regs)
 	for (i = 0; i < ARRAY_SIZE(regs->gpr); i++)
 		kvmppc_set_gpr(vcpu, i, regs->gpr[i]);
 
+	/*
+	 * If the hypercall decides to change register values
+	 * explicitly, we must ensure that we do not override
+	 * them upon return from that hypercall. Among others
+	 * this happens on system reset
+	 */
+	vcpu->arch.hcall_needed = 0;
+
 	return 0;
 }
 
diff --git a/arch/powerpc/kvm/powerpc.c b/arch/powerpc/kvm/powerpc.c
index 87f4dc8..75e196d 100644
--- a/arch/powerpc/kvm/powerpc.c
+++ b/arch/powerpc/kvm/powerpc.c
@@ -592,6 +592,11 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu, struct kvm_run *run)
 	} else if (vcpu->arch.hcall_needed) {
 		int i;
 
+		/*
+		 * Note: This might not be called if the hypervisor
+		 * call has done a set_regs(). In this case hcall_needed
+		 * is cleared. This is necessary for reset to work properly
+		 */
 		kvmppc_set_gpr(vcpu, 3, run->papr_hcall.ret);
 		for (i = 0; i < 9; ++i)
 			kvmppc_set_gpr(vcpu, 4 + i, run->papr_hcall.args[i]);

             reply	other threads:[~2012-07-27  6:18 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-07-27  6:18 Benjamin Herrenschmidt [this message]
2012-07-30 12:03 ` [PATCH] powerpc/kvm: Allow qemu hypercalls to set all regs Alexander Graf

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1343369904.2118.20.camel@pasglop \
    --to=benh@ozlabs.org \
    --cc=agraf@suse.de \
    --cc=kvm-ppc@vger.kernel.org \
    --cc=kvm@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).