mirrors/Bochs
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

- apply patch.smp-eio-readable-wli from William Lee Irwin III.

Browse Source 
My code did a panic if you tried to read the EOI register (the panic
  message was wrong but the concept was right).  However it turns out
  some OSes do actually read this register--hopefully they ignore the
  result.  So it should not panic.
This commit is contained in:
Bryce Denney 2002-03-20 02:51:47 +00:00
parent 180f62a303
commit 571ac50d1c
2 changed files with 6 additions and 78 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
bochs/cpu/apic.cc
View File

@ -1,5 +1,5 @@
/////////////////////////////////////////////////////////////////////////
// $Id: apic.cc,v 1.8 2001-10-03 13:10:37 bdenney Exp $
// $Id: apic.cc,v 1.9 2002-03-20 02:51:47 bdenney Exp $
/////////////////////////////////////////////////////////////////////////
//
#define NEED_CPU_REG_SHORTCUTS 1
@ -462,7 +462,11 @@ void bx_local_apic_c::read_aligned (Bit32u addr, Bit32u *data, unsigned len)
case 0xa0: // processor priority
*data = get_ppr (); break;
case 0xb0: // EOI
BX_PANIC(("EOI register not writable"));
/*
* Read-modify-write operations should operate without generating
* exceptions, and are used by some operating systems to EOI.
* The results of reads should be ignored by the OS.
*/
break;
case 0xd0: // logical destination
*data = (log_dest & 0xff) << 24; break;

76
bochs/patches/patch.smp-eoi-readable-wli
View File

@ -1,76 +0,0 @@
Date: Tue, 19 Mar 2002 17:52:41 -0800
From: William Lee Irwin III <wli@holomorphy.com>
To: bochs-developers@lists.sourceforge.net
Subject: [Bochs-developers] 03_ignore_eoi_read
Linux 2.4 uses the following to EOI in some situations:
static inline void ack_APIC_irq(void)
{
/*
* ack_APIC_irq() actually gets compiled as a single instruction:
* - a single rmw on Pentium/82489DX
* - a single write on P6+ cores (CONFIG_X86_GOOD_APIC)
* ... yummie.
*/
/* Docs say use 0 for future compatibility */
apic_write_around(APIC_EOI, 0);
}
Where apic_write_around() is defined as follows:
#ifdef CONFIG_X86_GOOD_APIC
# define FORCE_READ_AROUND_WRITE 0
# define apic_read_around(x)
# define apic_write_around(x,y) apic_write((x),(y))
#else
# define FORCE_READ_AROUND_WRITE 1
# define apic_read_around(x) apic_read(x)
# define apic_write_around(x,y) apic_write_atomic((x),(y))
#endif
After unraveling all this we have
static __inline void apic_write_atomic(unsigned long reg, unsigned long v)
{
xchg((volatile unsigned long *)(APIC_BASE+reg), v);
}
This is a read-modify-write operation.
Now BX_CPU_C::XCHG_EdGd(BxInstruction_t *i) eventually decides to do this:
read_RMW_virtual_dword(i->seg, i->rm_addr, &op1_32);
write_RMW_virtual_dword(op2_32);
which triggers the following in
void bx_local_apic_c::read_aligned (Bit32u addr, Bit32u *data, unsigned len):
case 0xb0: // EOI
BX_PANIC(("EOI register not writable"));
break;
which I assume is not the desired outcome.
What follows is a patch to ignore reads of the EOI register.
Cheers,
Bill
--- cpu/apic.cc.orig Tue Mar 19 17:35:21 2002
+++ cpu/apic.cc Tue Mar 19 17:37:57 2002
@@ -471,7 +471,11 @@
case 0xa0: // processor priority
*data = get_ppr (); break;
case 0xb0: // EOI
- BX_PANIC(("EOI register not writable"));
+ /*
+ * Read-modify-write operations should operate without generating
+ * exceptions, and are used by some operating systems to EOI.
+ * The results of reads should be ignored by the OS.
+ */
break;
case 0xd0: // logical destination
*data = (log_dest & 0xff) << 24; break;