Andi Kleen wrote:
> On Wednesday 01 November 2006 11:27, Rusty Russell wrote:
>
>> Create a paravirt.h header for all the critical operations which need
>> to be replaced with hypervisor calls, and include that instead of
>> defining native operations, when CONFIG_PARAVIRT.
>>
>
> Hmm, did this all ever compile in mainline? I had to do a few merges
> and in the end i get
>
> /home/lsrc/quilt/linux/kernel/spinlock.c: In function ?_spin_lock_irqsave?:
> include2/asm/spinlock.h:59: error: invalid 'asm': operand number missing after %
> -letter
> include2/asm/spinlock.h:59: error: invalid 'asm': operand number missing after %
> -letter
> include2/asm/spinlock.h:59: error: invalid 'asm': operand number missing after %
> -letter
> include2/asm/spinlock.h:59: error: invalid 'asm': operand number missing after %
> -letter
> include2/asm/spinlock.h:59: error: invalid 'asm': operand number missing after %
> -letter
> include2/asm/spinlock.h:59: error: invalid 'asm': operand number missing after %
> -letter
> include2/asm/spinlock.h:59: error: invalid 'asm': operand number missing after %
> -letter
> include2/asm/spinlock.h:59: error: invalid 'asm': operand number missing after %
> -letter
> {standard input}: Assembler messages:
> {standard input}:593: Error: undefined symbol `paravirt_ops' in operation
> {standard input}:593: Error: undefined symbol `PARAVIRT_irq_enable' in operation
> {standard input}:605: Error: undefined symbol `paravirt_ops' in operation
> {standard input}:605: Error: undefined symbol `PARAVIRT_irq_disable' in operatio
> n
>
Not seeing that here (on 2.6.19-rc2-mm2 with gcc 4.0.2).
> and lots of new warnings like
>
> /home/lsrc/quilt/linux/arch/i386/kernel/traps.c: In function ?set_intr_gate?:
> /home/lsrc/quilt/linux/arch/i386/kernel/traps.c:1165: warning: implicit declarat
> ion of function ?_set_gate?
> /home/lsrc/quilt/linux/arch/i386/kernel/cpu/common.c: In function ?_cpu_init?:
> /home/lsrc/quilt/linux/arch/i386/kernel/cpu/common.c:754: warning: implicit decl
> aration of function ?__set_tss_desc'
>
Sounds like desc.h got reordered. Somewhere, there was a broken patch
once that did this, I thought we fixed that.
> /home/lsrc/quilt/linux/arch/i386/kernel/cpu/mcheck/p4.c: In function ?intel_mach
> ine_check?:
> /home/lsrc/quilt/linux/arch/i386/kernel/cpu/mcheck/p4.c:158: warning: ?dbg.eax?
> may be used uninitialized in this function
> /home/lsrc/quilt/linux/arch/i386/kernel/cpu/mcheck/p4.c:158: warning: ?dbg.ebx?
> may be used uninitialized in this function
> /home/lsrc/quilt/linux/arch/i386/kernel/cpu/mcheck/p4.c:158: warning: ?dbg.ecx?
> may be used uninitialized in this function
> /home/lsrc/quilt/linux/arch/i386/kernel/cpu/mcheck/p4.c:158: warning: ?dbg.edx?
> may be used uninitialized in this function
> /home/lsrc/quilt/linux/arch/i386/kernel/cpu/mcheck/p4.c:158: warning: ?dbg.esi?
> may be used uninitialized in this function
> /home/lsrc/quilt/linux/arch/i386/kernel/cpu/mcheck/p4.c:158: warning: ?dbg.edi?
> may be used uninitialized in this function
> /home/lsrc/quilt/linux/arch/i386/kernel/cpu/mcheck/p4.c:158: warning: ?dbg.ebp?
> may be used uninitialized in this function
> /home/lsrc/quilt/linux/arch/i386/kernel/cpu/mcheck/p4.c:158: warning: ?dbg.esp?
> may be used uninitialized in this function
> /home/lsrc/quilt/linux/arch/i386/kernel/cpu/mcheck/p4.c:158: warning: ?dbg.eflag
> s? may be used uninitialized in this function
> /home/lsrc/quilt/linux/arch/i386/kernel/cpu/mcheck/p4.c:158: warning: ?dbg.eip?
> may be used uninitialized in this function
>
Those appear to be valid warnings, with or without paravirt, due to the
tacky glued inline oddity of intel_get_extended_msrs.
Zach
