From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43)
	id 1NuWww-0002as-1E
	for qemu-devel@nongnu.org; Wed, 24 Mar 2010 16:13:10 -0400
Received: from [140.186.70.92] (port=36600 helo=eggs.gnu.org)
	by lists.gnu.org with esmtp (Exim 4.43) id 1NuWwf-0002aB-Dn
	for qemu-devel@nongnu.org; Wed, 24 Mar 2010 16:13:09 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.69)
	(envelope-from <rth@twiddle.net>) id 1NuWvw-0004nL-9u
	for qemu-devel@nongnu.org; Wed, 24 Mar 2010 16:12:13 -0400
Received: from are.twiddle.net ([75.149.56.221]:49745)
	by eggs.gnu.org with esmtp (Exim 4.69)
	(envelope-from <rth@twiddle.net>) id 1NuWvw-0004nB-30
	for qemu-devel@nongnu.org; Wed, 24 Mar 2010 16:12:08 -0400
Message-ID: <4BAA7216.3020105@twiddle.net>
Date: Wed, 24 Mar 2010 13:12:06 -0700
From: Richard Henderson <rth@twiddle.net>
MIME-Version: 1.0
Subject: Re: [Qemu-devel] Re: Compile files only once: some planning
References: <f43fc5581003231443x5492dd7eve541fd8104170a60@mail.gmail.com>	<4BA9DFAA.3070107@redhat.com>
	<4BA9F52C.6070309@twiddle.net>	<4BAA25A7.7000201@redhat.com>
	<4BAA46D8.1080505@twiddle.net>
In-Reply-To: <4BAA46D8.1080505@twiddle.net>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
List-Id: qemu-devel.nongnu.org
List-Unsubscribe: <http://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.gnu.org/pipermail/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <http://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: Paolo Bonzini <pbonzini@redhat.com>
Cc: Blue Swirl <blauwirbel@gmail.com>, qemu-devel <qemu-devel@nongnu.org>

On 03/24/2010 10:07 AM, Richard Henderson wrote:
> struct CPUSmallCommonState
> {
>     // most of the stuff from CPU_COMMON.
>     // sorted for some thought of padding elimination.  ;-)
> };
> 
> struct CPULargeCommonState
> {
>     CPUTLBEntry tlb_table[NB_MMU_MODES][CPU_TLB_SIZE];
>     target_phys_addr_t iotlb[NB_MMU_MODES][CPU_TLB_SIZE];
>     struct TranslationBlock *tb_jmp_cache[TB_JMP_CACHE_SIZE];
>     jmp_buf jmp_env;
> };
...
> Now.  If you're compiling a file for which cpu-specific code is ok:
> 
> register CPUXYZLargeState *env __asm__(AREG0);
> #define ENV_SMALL_COMMON_STATE    (&env->s.common_s)
> #define ENV_LARGE_COMMON_STATE    (&env->common_l)
> 
> If you're compiling a file which is supposed to be independant of cpu:
> 
> register CPUSmallCommonState *env __asm__(AREG0);
> #define ENV_SMALL_COMMON_STATE    (env)
> #define ENV_LARGE_COMMON_STATE    ((CPULargeCommonState *)((char *)env + cpu_large_state_offset))

On 03/24/2010 11:00 AM, Blue Swirl wrote:
> One trick is to define a fixed offset (about half CPUState size) and
> make env point to CPUState + offset. Then the negative part of the
> offset space can be used efficiently. But this just doubles the space
> that can be accessed fast, so it's not a big win.

A good idea.  We can eliminate the cpu_large_state_offset from above via:

struct CPUSmallCommonState
{
    // most of the stuff from CPU_COMMON.
} __attribute__((aligned));

struct CPUXYZPrivateState
{
    // all the cpu-specific stuff
};

struct CPUXYZSmallState
{
    CPUXYZPrivateState p;
    CPUSmallCommonState s;
};

struct CPUXYZAllState
{
    CPUXYZSmallState s;
    CPULargeCommonState l;  // ARG0 register points here.
};

register void *biased_env __asm__(AREG0);

static inline CPUXYZPrivateState *get_env_cpu_private(void)
{
    return &((CPUXYZSmallState *)biased_env - 1)->p;
}

static inline CPUSmallCommonState *get_env_common_small(void)
{
    return (CPUSmallCommonState *)biased_env - 1;
}

static inline CPULargeCommonState *get_env_common_large(void)
{
    return (CPULargeCommonState *)biased_env;
}


r~