From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([208.118.235.92]:53834)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <kwolf@redhat.com>) id 1UFjps-0004kf-Qq
	for qemu-devel@nongnu.org; Wed, 13 Mar 2013 07:27:16 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <kwolf@redhat.com>) id 1UFjpm-00021o-KX
	for qemu-devel@nongnu.org; Wed, 13 Mar 2013 07:27:08 -0400
Received: from mx1.redhat.com ([209.132.183.28]:27489)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <kwolf@redhat.com>) id 1UFjpm-00021a-Bf
	for qemu-devel@nongnu.org; Wed, 13 Mar 2013 07:27:02 -0400
Date: Wed, 13 Mar 2013 12:26:59 +0100
From: Kevin Wolf <kwolf@redhat.com>
Message-ID: <20130313112659.GE2309@dhcp-200-207.str.redhat.com>
References: <1362491612-19226-1-git-send-email-aliguori@us.ibm.com>
	<1362491612-19226-2-git-send-email-aliguori@us.ibm.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <1362491612-19226-2-git-send-email-aliguori@us.ibm.com>
Subject: Re: [Qemu-devel] [RFC PATCH 1/8] qtest: add libqos
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: Anthony Liguori <aliguori@us.ibm.com>
Cc: qemu-devel@nongnu.org, Stefan Hajnoczi <stefanha@redhat.com>

Am 05.03.2013 um 14:53 hat Anthony Liguori geschrieben:
> This includes basic PCI support.
> 
> Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>

> +static void *qpci_pc_iomap(QPCIBus *bus, QPCIDevice *dev, int barno)
> +{
> +    QPCIBusPC *s = container_of(bus, QPCIBusPC, bus);
> +    static const int bar_reg_map[] = {
> +        PCI_BASE_ADDRESS_0, PCI_BASE_ADDRESS_1, PCI_BASE_ADDRESS_2,
> +        PCI_BASE_ADDRESS_3, PCI_BASE_ADDRESS_4, PCI_BASE_ADDRESS_5,
> +    };
> +    int bar_reg;
> +    uint32_t addr;
> +    uint64_t size;
> +
> +    g_assert(barno >= 0 && barno <= 5);
> +    bar_reg = bar_reg_map[barno];
> +
> +    qpci_config_writel(dev, bar_reg, 0xFFFFFFFF);
> +    addr = qpci_config_readl(dev, bar_reg);
> +
> +    size = (1ULL << ctol(addr));

This doesn't look right. It should be something like:

    size = (1ULL << ctzl(addr & ~0x3));

In fact, what must be masked out differs for I/O (0x3) and memory (0xf).

> --- /dev/null
> +++ b/tests/libqos/pci.c
> @@ -0,0 +1,108 @@
> +#include "libqos/pci.h"
> +
> +#include "hw/pci/pci_regs.h"
> +#include <glib.h>
> +
> +#include <stdio.h>
> +
> +QPCIDevice *qpci_device_find(QPCIBus *bus, int devfn)
> +{
> +    QPCIDevice *dev;
> +
> +    dev = g_malloc0(sizeof(*dev));

Where is the matching free? I can't seem to destroy a device I
once queried.

> +    dev->bus = bus;
> +    dev->devfn = devfn;
> +
> +    if (qpci_config_readw(dev, PCI_VENDOR_ID) == 0xFFFF) {
> +        printf("vendor id is %x\n", qpci_config_readw(dev, PCI_VENDOR_ID));
> +        g_free(dev);
> +        return NULL;
> +    }
> +
> +    return dev;
> +}
> +
> +void qpci_device_enable(QPCIDevice *dev)
> +{
> +    uint16_t cmd;
> +
> +    /* FIXME -- does this need to be a bus callout? */
> +    cmd = qpci_config_readw(dev, PCI_COMMAND);
> +    cmd |= PCI_COMMAND_IO | PCI_COMMAND_MEMORY;
> +    qpci_config_writew(dev, PCI_COMMAND, cmd);
> +}

Wouldn't it make sense to enable bus mastering here as well? Forgetting
to do this manually is a trap that's easy to fall in...

Kevin