qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Jan Kiszka <jan.kiszka@web.de>
To: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Cc: qemu-devel@nongnu.org
Subject: [Qemu-devel] Re: Breakage with local APIC routing
Date: Wed, 13 Aug 2008 11:08:59 +0200	[thread overview]
Message-ID: <48A2A4AB.6050601@web.de> (raw)
In-Reply-To: <48A297F0.9000700@web.de>

Jan Kiszka wrote:
> Johannes Schindelin wrote:
>> Hi,
>>
>> due to the change in revision 3371 (well, at that time, CVS was used, 
>> which was no better than Subversion) installation of win64 is broken in 
>> QEmu.  The commit message reads like this:
>>
>> 	Don't route PIC interrupts through the local APIC if the local 
>> 	APIC config says so. By Ari Kivity.
>>
>> A bit of research showed that the patch was actually originally from Qing 
>> He, but he told me privately that the part that actually broke win64 (the 
>> removal of the call to cpu_reset_interrupt(), as opposed to moving that 
>> call into the "else" condition) was not part of his patch.
>>
>> Unfortunately, a lot has been done to the APIC handling in the meantime, 
>> so it is not a simple matter of a revert.
>>
>> Being a complete idiot when it comes to APICs, I have no clue how to fix 
>> the issue.
>>
>> However, I am quite willing to test whatever patch is thrown at me.
>>
>> Can somebody help?
> 
> I recalled some earlier post on this which claimed to fix the issue and
> found it in the archive:
> 
> http://permalink.gmane.org/gmane.comp.emulators.qemu/25415
> 
> However, no one replied on this, and I'm right now not sure if the fix
> is OK for all cases. But it is a starting point for new discussion about
> what is actually borken here.

This might not be the last word, but my feeling that it's closer to the
actual issue than the older fix. Please give it a try. Comments welcome.

--------

Ensure that PIC-delivered IRQs are properly de-asserted in case the APIC
is in EXTINT or FIXED mode (with level-triggering selected) on LINT0.
Should fix Win64 boot issues. This patch also cleans up a bit the
interface between PIC and APIC, making apic_local_deliver private again.

Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
---
 hw/apic.c |   23 ++++++++++++++++++++++-
 hw/pc.c   |    5 +----
 hw/pc.h   |    4 +---
 3 files changed, 24 insertions(+), 8 deletions(-)

Index: b/hw/apic.c
===================================================================
--- a/hw/apic.c
+++ b/hw/apic.c
@@ -166,7 +166,7 @@ static inline void reset_bit(uint32_t *t
     tab[i] &= ~mask;
 }
 
-void apic_local_deliver(CPUState *env, int vector)
+static void apic_local_deliver(CPUState *env, int vector)
 {
     APICState *s = env->apic_state;
     uint32_t lvt = s->lvt[vector];
@@ -197,6 +197,27 @@ void apic_local_deliver(CPUState *env, i
     }
 }
 
+void apic_deliver_pic_intr(CPUState *env, int level)
+{
+    if (level)
+        apic_local_deliver(env, APIC_LVT_LINT0);
+    else {
+        APICState *s = env->apic_state;
+        uint32_t lvt = s->lvt[APIC_LVT_LINT0];
+
+        switch ((lvt >> 8) & 7) {
+        case APIC_DM_FIXED:
+            if (!(lvt & APIC_LVT_LEVEL_TRIGGER))
+                break;
+            reset_bit(s->irr, lvt & 0xff);
+            /* fall through */
+        case APIC_DM_EXTINT:
+            cpu_reset_interrupt(env, CPU_INTERRUPT_HARD);
+            break;
+        }
+    }
+}
+
 #define foreach_apic(apic, deliver_bitmask, code) \
 {\
     int __i, __j, __mask;\
Index: b/hw/pc.c
===================================================================
--- a/hw/pc.c
+++ b/hw/pc.c
@@ -118,12 +118,9 @@ static void pic_irq_request(void *opaque
 {
     CPUState *env = first_cpu;
 
-    if (!level)
-        return;
-
     while (env) {
         if (apic_accept_pic_intr(env))
-            apic_local_deliver(env, APIC_LINT0);
+            apic_deliver_pic_intr(env, level);
         env = env->next_cpu;
     }
 }
Index: b/hw/pc.h
===================================================================
--- a/hw/pc.h
+++ b/hw/pc.h
@@ -40,11 +40,9 @@ void irq_info(void);
 /* APIC */
 typedef struct IOAPICState IOAPICState;
 
-#define APIC_LINT0	3
-
 int apic_init(CPUState *env);
 int apic_accept_pic_intr(CPUState *env);
-void apic_local_deliver(CPUState *env, int vector);
+void apic_deliver_pic_intr(CPUState *env, int level);
 int apic_get_interrupt(CPUState *env);
 IOAPICState *ioapic_init(void);
 void ioapic_set_irq(void *opaque, int vector, int level);

  reply	other threads:[~2008-08-13  9:09 UTC|newest]

Thread overview: 46+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-08-11 14:28 [Qemu-devel] Breakage with local APIC routing Johannes Schindelin
2008-08-13  8:14 ` [Qemu-devel] " Jan Kiszka
2008-08-13  9:08   ` Jan Kiszka [this message]
2008-08-17 15:00     ` Johannes Schindelin
2008-08-17 15:28   ` Johannes Schindelin
2008-08-17 16:52     ` Jan Kiszka
2009-08-13 16:21       ` Johannes Schindelin
2009-08-25  6:56         ` Jan Kiszka
2009-08-25  8:24           ` Mohammed Gamal
2009-08-25  8:59             ` Jan Kiszka
2009-08-25  9:09               ` Mohammed Gamal
2009-08-25  9:33                 ` Jan Kiszka
2009-08-25 10:16                   ` Mohammed Gamal
2009-08-25 10:21                     ` Mohammed Gamal
2009-08-25 10:38                       ` Jan Kiszka
2009-08-26 22:10                         ` Juergen Lock
2009-08-27 17:56                           ` Jan Kiszka
2009-08-29 21:18                             ` Juergen Lock
2009-08-31  7:47                               ` Jan Kiszka
2009-08-31 21:27                                 ` Juergen Lock
2009-09-01 20:12                                   ` Juergen Lock
2009-09-07 20:59                                     ` FreeBSD timing issues and qemu (was: Re: [Qemu-devel] Re: Breakage with local APIC routing) Juergen Lock
2009-09-08  2:17                                       ` Ryan Stone
2009-09-09 20:15                                         ` Juergen Lock
2009-09-09 20:39                                           ` Ryan Stone
2009-09-09 20:46                                       ` Luigi Rizzo
2009-09-10 17:46                                         ` Juergen Lock
2009-09-10 19:08                                           ` Luigi Rizzo
2009-09-10 20:44                                             ` Juergen Lock
2009-09-11 15:22                                             ` John Baldwin
2009-09-11 17:03                                               ` Luigi Rizzo
2009-09-11 17:01                                                 ` John Baldwin
2009-09-12 15:48                                                   ` Luigi Rizzo
2009-08-25 11:48           ` [Qemu-devel] Re: Breakage with local APIC routing Johannes Schindelin
2009-08-25 12:31             ` Jan Kiszka
2009-08-25 13:36               ` Johannes Schindelin
2009-08-26  7:31                 ` Jan Kiszka
2009-08-26  9:27                   ` Avi Kivity
2009-08-27 18:07                 ` Jan Kiszka
2008-08-17 17:32 ` [Qemu-devel] " Stefan Weil
2008-08-17 18:58   ` [Qemu-devel] " Jan Kiszka
2008-08-17 20:40     ` Stefan Weil
2008-08-17 20:23   ` [Qemu-devel] " Aurelien Jarno
2008-08-17 21:10     ` Stefan Weil
2008-08-18 14:27       ` Anthony Liguori
2008-08-17 21:14     ` [Qemu-devel] " Jan Kiszka

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=48A2A4AB.6050601@web.de \
    --to=jan.kiszka@web.de \
    --cc=Johannes.Schindelin@gmx.de \
    --cc=qemu-devel@nongnu.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).