From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:54042)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <paolo.bonzini@gmail.com>) id 1aHZMl-0000fw-C9
	for qemu-devel@nongnu.org; Fri, 08 Jan 2016 10:54:16 -0500
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <paolo.bonzini@gmail.com>) id 1aHZMi-0006bX-1x
	for qemu-devel@nongnu.org; Fri, 08 Jan 2016 10:54:15 -0500
Received: from mail-wm0-x22c.google.com ([2a00:1450:400c:c09::22c]:33810)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <paolo.bonzini@gmail.com>) id 1aHZMh-0006bQ-RR
	for qemu-devel@nongnu.org; Fri, 08 Jan 2016 10:54:11 -0500
Received: by mail-wm0-x22c.google.com with SMTP id u188so142044594wmu.1
	for <qemu-devel@nongnu.org>; Fri, 08 Jan 2016 07:54:11 -0800 (PST)
Sender: Paolo Bonzini <paolo.bonzini@gmail.com>
References: <1452265115-27177-1-git-send-email-markmb@redhat.com>
From: Paolo Bonzini <pbonzini@redhat.com>
Message-ID: <568FDB9F.7080709@redhat.com>
Date: Fri, 8 Jan 2016 16:54:07 +0100
MIME-Version: 1.0
In-Reply-To: <1452265115-27177-1-git-send-email-markmb@redhat.com>
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 8bit
Subject: Re: [Qemu-devel] [PATCH] Add optionrom compatible with fw_cfg DMA
	version
List-Id: <qemu-devel.nongnu.org>
List-Unsubscribe: <https://lists.nongnu.org/mailman/options/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.nongnu.org/archive/html/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <https://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: =?UTF-8?Q?Marc_Mar=c3=ad?= <markmb@redhat.com>, qemu-devel <qemu-devel@nongnu.org>, "Gabriel L. Somlo" <gsomlo@gmail.com>, Kevin O'Connor <kevin@koconnor.net>, Laszlo Ersek <lersek@redhat.com>, Stefan Hajnoczi <stefanha@redhat.com>



On 08/01/2016 15:58, Marc Marí wrote:
> 
> +static inline uint16_t readw_addr32(const void *addr) {
> +    uint16_t val;
> +    asm("addr32 movw %1, %0" : "=r"(val) : "g"(addr));
> +    barrier();
> +    return val;
> +}
> +

Does SeaBIOS ensure that DS base is zero here?

> +static void transition32(void)
> +{
> +    extern void *gdt;
> +    uint32_t data_segment;
> +    struct length_addr rombios_gdt;
> +
> +    data_segment = read_ds();
> +    rombios_gdt.addr = (uint32_t)((data_segment << 4) + (uint32_t)(&gdt));
> +    rombios_gdt.length = (3 * 8) - 1;
> +
> +    /* Load GDT */
> +    asm("data32 lgdt %0" : : "m"(rombios_gdt): "memory");
> +
> +	/* Get us to protected mode and set ES to a 32 bit segment */
> +    asm("mov $1, %%eax\n"
> +        "mov %%eax, %%cr0\n"
> +        "mov $0x10, %%eax\n"
> +        "mov %%eax, %%es\n"
> +        : : : "eax");
> +
> +    /* We're now running in 16-bit CS, but 32-bit ES! */
> +}

Do you even need to go to 32-bit mode?  The only reason to do so in the
original ROM was to do a "rep insb" above 1 GB, but here fw_cfg can do
DMA to high addresses for you.

Paolo