From: Don Dutile on
stefano(a)stabellini.net wrote:
> From: Stefano Stabellini <stefano.stabellini(a)eu.citrix.com>
>
> Add a xen_emul_unplug command line option to the kernel to unplug
> xen emulated disks and nics.
>
> Set the default value of xen_emul_unplug depending on whether or
> not the Xen PV frontends and the Xen platform PCI driver have
> been compiled for this kernel (modules or built-in are both OK).
>
> Signed-off-by: Stefano Stabellini <stefano.stabellini(a)eu.citrix.com>
> ---
> Documentation/kernel-parameters.txt | 11 +++
> arch/x86/xen/Makefile | 2 +-
> arch/x86/xen/enlighten.c | 1 +
> arch/x86/xen/platform-pci-unplug.c | 130 +++++++++++++++++++++++++++++++++++
> arch/x86/xen/xen-ops.h | 1 +
> drivers/xen/platform-pci.c | 4 +
> drivers/xen/xenbus/xenbus_probe.c | 4 +
> include/xen/platform_pci.h | 49 +++++++++++++
> 8 files changed, 201 insertions(+), 1 deletions(-)
> create mode 100644 arch/x86/xen/platform-pci-unplug.c
> create mode 100644 include/xen/platform_pci.h
>
> diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
> index 839b21b..716eea8 100644
> --- a/Documentation/kernel-parameters.txt
> +++ b/Documentation/kernel-parameters.txt
> @@ -113,6 +113,7 @@ parameter is applicable:
> More X86-64 boot options can be found in
> Documentation/x86/x86_64/boot-options.txt .
> X86 Either 32bit or 64bit x86 (same as X86-32+X86-64)
> + XEN Xen support is enabled
>
> In addition, the following text indicates that the option:
>
> @@ -2834,6 +2835,16 @@ and is between 256 and 4096 characters. It is defined in the file
> xd= [HW,XT] Original XT pre-IDE (RLL encoded) disks.
> xd_geo= See header of drivers/block/xd.c.
>
> + xen_emul_unplug= [HW,X86,XEN]
> + Unplug Xen emulated devices
> + Format: [unplug0,][unplug1]
> + ide-disks -- unplug primary master IDE devices
> + aux-ide-disks -- unplug non-primary-master IDE devices
> + nics -- unplug network devices
> + all -- unplug all emulated devices (NICs and IDE disks)
> + ignore -- continue loading the Xen platform PCI driver even
> + if the version check failed
> +
> xirc2ps_cs= [NET,PCMCIA]
> Format:
> <irq>,<irq_mask>,<io>,<full_duplex>,<do_sound>,<lockup_hack>[,<irq2>[,<irq3>[,<irq4>]]]
> diff --git a/arch/x86/xen/Makefile b/arch/x86/xen/Makefile
> index 3bb4fc2..9309546 100644
> --- a/arch/x86/xen/Makefile
> +++ b/arch/x86/xen/Makefile
> @@ -12,7 +12,7 @@ CFLAGS_mmu.o := $(nostackp)
>
> obj-y := enlighten.o setup.o multicalls.o mmu.o irq.o \
> time.o xen-asm.o xen-asm_$(BITS).o \
> - grant-table.o suspend.o
> + grant-table.o suspend.o platform-pci-unplug.o
>
> obj-$(CONFIG_SMP) += smp.o
> obj-$(CONFIG_PARAVIRT_SPINLOCKS)+= spinlock.o
> diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
> index bcd98a9..5126b53 100644
> --- a/arch/x86/xen/enlighten.c
> +++ b/arch/x86/xen/enlighten.c
> @@ -1317,6 +1317,7 @@ void __init xen_hvm_guest_init(void)
> if (xen_feature(XENFEAT_hvm_callback_vector))
> xen_have_vector_callback = 1;
> register_cpu_notifier(&xen_hvm_cpu_notifier);
> + xen_unplug_emulated_devices();
> have_vcpu_info_placement = 0;
> x86_init.irqs.intr_init = xen_init_IRQ;
> xen_hvm_init_time_ops();
> diff --git a/arch/x86/xen/platform-pci-unplug.c b/arch/x86/xen/platform-pci-unplug.c
> new file mode 100644
> index 0000000..d686440
> --- /dev/null
> +++ b/arch/x86/xen/platform-pci-unplug.c
> @@ -0,0 +1,130 @@
> +/******************************************************************************
> + * platform-pci-unplug.c
> + *
> + * Xen platform PCI device driver
> + * Copyright (c) 2010, Citrix
> + *
> + * This program is free software; you can redistribute it and/or modify it
> + * under the terms and conditions of the GNU General Public License,
> + * version 2, as published by the Free Software Foundation.
> + *
> + * This program is distributed in the hope it will be useful, but WITHOUT
> + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
> + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
> + * more details.
> + *
> + * You should have received a copy of the GNU General Public License along with
> + * this program; if not, write to the Free Software Foundation, Inc., 59 Temple
> + * Place - Suite 330, Boston, MA 02111-1307 USA.
> + *
> + */
> +
> +#include <asm/io.h>
> +
> +#include <linux/init.h>
> +#include <linux/module.h>
> +
> +#include <xen/platform_pci.h>
> +
> +/* boolean to signal that the platform pci device can be used */
> +bool xen_platform_pci_enabled;
> +EXPORT_SYMBOL_GPL(xen_platform_pci_enabled);
> +static int xen_emul_unplug;
> +
> +static int __init check_platform_magic(void)
> +{
> + short magic;
> + char protocol;
> +
> + magic = inw(XEN_IOPORT_MAGIC);
> + if (magic != XEN_IOPORT_MAGIC_VAL) {
> + printk(KERN_ERR "Xen Platform PCI: unrecognised magic value\n");
> + return -1;
> + }
> +
> + protocol = inb(XEN_IOPORT_PROTOVER);
> +
> + printk(KERN_DEBUG "Xen Platform PCI: I/O protocol version %d\n",
> + protocol);
> +
> + switch (protocol) {
> + case 1:
> + outw(XEN_IOPORT_LINUX_PRODNUM, XEN_IOPORT_PRODNUM);
> + outl(XEN_IOPORT_LINUX_DRVVER, XEN_IOPORT_DRVVER);
> + if (inw(XEN_IOPORT_MAGIC) != XEN_IOPORT_MAGIC_VAL) {
> + printk(KERN_ERR "Xen Platform: blacklisted by host\n");
> + return -3;
> + }
> + break;
> + default:
> + printk(KERN_WARNING "Xen Platform PCI: unknown I/O protocol version");
> + return -2;
> + }
> +
> + return 0;
> +}
> +
> +void __init xen_unplug_emulated_devices(void)
> +{
> + int r;
> +
> + /* check the version of the xen platform PCI device */
> + r = check_platform_magic();
> + /* If the version matches enable the Xen platform PCI driver.
> + * Also enable the Xen platform PCI driver if the version is really old
> + * and the user told us to ignore it. */
> + if (!r || (r == -1 && (xen_emul_unplug & XEN_UNPLUG_IGNORE)))
> + xen_platform_pci_enabled = 1;
> + /* Set the default value of xen_emul_unplug depending on whether or
> + * not the Xen PV frontends and the Xen platform PCI driver have
> + * been compiled for this kernel (modules or built-in are both OK). */
> + if (xen_platform_pci_enabled && !xen_emul_unplug) {
> + if (xen_must_unplug_nics()) {
> + printk(KERN_INFO "Netfront and the Xen platform PCI driver have "
> + "been compiled for this kernel: unplug emulated NICs.\n");
> + xen_emul_unplug |= XEN_UNPLUG_ALL_NICS;
> + }
> + if (xen_must_unplug_disks()) {
> + printk(KERN_INFO "Blkfront and the Xen platform PCI driver have "
> + "been compiled for this kernel: unplug emulated disks.\n"
> + "You might have to change the root device\n"
> + "from /dev/hd[a-d] to /dev/xvd[a-d]\n"
> + "in your root= kernel command line option\n");
> + xen_emul_unplug |= XEN_UNPLUG_ALL_IDE_DISKS;
> + }
> + }
> + /* Now unplug the emulated devices */
> + if (xen_platform_pci_enabled && !(xen_emul_unplug & XEN_UNPLUG_IGNORE))
> + outw(xen_emul_unplug, XEN_IOPORT_UNPLUG);
> +}
> +
> +static int __init parse_xen_emul_unplug(char *arg)
> +{
> + char *p, *q;
> + int l;
> +
> + for (p = arg; p; p = q) {
> + q = strchr(p, ',');
> + if (q) {
> + l = q - p;
> + q++;
> + } else {
> + l = strlen(p);
> + }
> + if (!strncmp(p, "all", l))
> + xen_emul_unplug |= XEN_UNPLUG_ALL;
> + else if (!strncmp(p, "ide-disks", l))
> + xen_emul_unplug |= XEN_UNPLUG_ALL_IDE_DISKS;
> + else if (!strncmp(p, "aux-ide-disks", l))
> + xen_emul_unplug |= XEN_UNPLUG_AUX_IDE_DISKS;
> + else if (!strncmp(p, "nics", l))
> + xen_emul_unplug |= XEN_UNPLUG_ALL_NICS;
> + else if (!strncmp(p, "ignore", l))
> + xen_emul_unplug |= XEN_UNPLUG_IGNORE;
> + else
> + printk(KERN_WARNING "unrecognised option '%s' "
> + "in module parameter 'dev_unplug'\n", p);
> + }
> + return 0;
> +}
> +early_param("xen_emul_unplug", parse_xen_emul_unplug);
> diff --git a/arch/x86/xen/xen-ops.h b/arch/x86/xen/xen-ops.h
> index 089d189..ed77694 100644
> --- a/arch/x86/xen/xen-ops.h
> +++ b/arch/x86/xen/xen-ops.h
> @@ -40,6 +40,7 @@ void xen_vcpu_restore(void);
>
> void xen_callback_vector(void);
> void xen_hvm_init_shared_info(void);
> +void __init xen_unplug_emulated_devices(void);
>
> void __init xen_build_dynamic_phys_to_machine(void);
>
> diff --git a/drivers/xen/platform-pci.c b/drivers/xen/platform-pci.c
> index 10b92ec..35f162d 100644
> --- a/drivers/xen/platform-pci.c
> +++ b/drivers/xen/platform-pci.c
> @@ -27,6 +27,7 @@
> #include <linux/module.h>
> #include <linux/pci.h>
>
> +#include <xen/platform_pci.h>
> #include <xen/grant_table.h>
> #include <xen/xenbus.h>
> #include <xen/events.h>
> @@ -195,6 +196,9 @@ static int __init platform_pci_module_init(void)
> {
> int rc;
>
> + if (!xen_platform_pci_enabled)
> + return -ENODEV;
> +

The problem with this check/enable is that if you run
this on an older qemu-xen that doesn't have unplug support,
it fails pv-hvm configuration.

But, all that means is that you can't use an xvd as the boot device,
and you have to use an emulated IDE device as boot device.
There are a couple ways to configure the vnif correctly (in guest
or in xen guest config file).

So, on an older (say, rhel5) xen, I don't have this check;
the boot device is required to be spec'd as hda, not vda, and
xen-blkfront is not allowed to configure blk major nums
for IDE (& SCSI) (to avoid 2 drivers twiddling w/same phys backend... not good!).

When that older qemu is updated, this restriction will be lifted
(in the guest kernel & related xen guest config file spec).

> rc = pci_register_driver(&platform_driver);
> if (rc) {
> printk(KERN_INFO DRV_NAME
> diff --git a/drivers/xen/xenbus/xenbus_probe.c b/drivers/xen/xenbus/xenbus_probe.c
> index 3d941ec..243279a 100644
> --- a/drivers/xen/xenbus/xenbus_probe.c
> +++ b/drivers/xen/xenbus/xenbus_probe.c
> @@ -56,6 +56,7 @@
> #include <xen/events.h>
> #include <xen/page.h>
>
> +#include <xen/platform_pci.h>
> #include <xen/hvm.h>
>
> #include "xenbus_comms.h"
> @@ -977,6 +978,9 @@ static void wait_for_devices(struct xenbus_driver *xendrv)
> #ifndef MODULE
> static int __init boot_wait_for_devices(void)
> {
> + if (xen_hvm_domain() && !xen_platform_pci_enabled)
> + return -ENODEV;
> +

This check also fails on qemu-xen that doesn't support
ide-unplug but pv-hvm is configured in (aka, on rhel5 today).
The flag should be something different.


> ready_to_wait_for_devices = 1;
> wait_for_devices(NULL);
> return 0;
> diff --git a/include/xen/platform_pci.h b/include/xen/platform_pci.h
> new file mode 100644
> index 0000000..afa8855
> --- /dev/null
> +++ b/include/xen/platform_pci.h
> @@ -0,0 +1,49 @@
> +#ifndef _XEN_PLATFORM_PCI_H
> +#define _XEN_PLATFORM_PCI_H
> +
> +#define XEN_IOPORT_MAGIC_VAL 0x49d2
> +#define XEN_IOPORT_LINUX_PRODNUM 0x0003
> +#define XEN_IOPORT_LINUX_DRVVER 0x0001
> +
> +#define XEN_IOPORT_BASE 0x10
> +
> +#define XEN_IOPORT_PLATFLAGS (XEN_IOPORT_BASE + 0) /* 1 byte access (R/W) */
> +#define XEN_IOPORT_MAGIC (XEN_IOPORT_BASE + 0) /* 2 byte access (R) */
> +#define XEN_IOPORT_UNPLUG (XEN_IOPORT_BASE + 0) /* 2 byte access (W) */
> +#define XEN_IOPORT_DRVVER (XEN_IOPORT_BASE + 0) /* 4 byte access (W) */
> +
> +#define XEN_IOPORT_SYSLOG (XEN_IOPORT_BASE + 2) /* 1 byte access (W) */
> +#define XEN_IOPORT_PROTOVER (XEN_IOPORT_BASE + 2) /* 1 byte access (R) */
> +#define XEN_IOPORT_PRODNUM (XEN_IOPORT_BASE + 2) /* 2 byte access (W) */
> +
> +#define XEN_UNPLUG_ALL_IDE_DISKS 1
> +#define XEN_UNPLUG_ALL_NICS 2
> +#define XEN_UNPLUG_AUX_IDE_DISKS 4
> +#define XEN_UNPLUG_ALL 7
> +#define XEN_UNPLUG_IGNORE 8
> +
> +static inline int xen_must_unplug_nics(void) {
> +#if (defined(CONFIG_XEN_NETDEV_FRONTEND) || \
> + defined(CONFIG_XEN_NETDEV_FRONTEND_MODULE)) && \
> + (defined(CONFIG_XEN_PLATFORM_PCI) || \
> + defined(CONFIG_XEN_PLATFORM_PCI_MODULE))
> + return 1;
> +#else
> + return 0;
> +#endif
> +}
> +
> +static inline int xen_must_unplug_disks(void) {
> +#if (defined(CONFIG_XEN_BLKDEV_FRONTEND) || \
> + defined(CONFIG_XEN_BLKDEV_FRONTEND_MODULE)) && \
> + (defined(CONFIG_XEN_PLATFORM_PCI) || \
> + defined(CONFIG_XEN_PLATFORM_PCI_MODULE))
> + return 1;
> +#else
> + return 0;
> +#endif
> +}
> +
> +extern bool xen_platform_pci_enabled;
> +
> +#endif /* _XEN_PLATFORM_PCI_H */

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Stefano Stabellini on
On Thu, 1 Jul 2010, Don Dutile wrote:
> The problem with this check/enable is that if you run
> this on an older qemu-xen that doesn't have unplug support,
> it fails pv-hvm configuration.
>
> But, all that means is that you can't use an xvd as the boot device,
> and you have to use an emulated IDE device as boot device.
> There are a couple ways to configure the vnif correctly (in guest
> or in xen guest config file).
>
> So, on an older (say, rhel5) xen, I don't have this check;
> the boot device is required to be spec'd as hda, not vda, and
> xen-blkfront is not allowed to configure blk major nums
> for IDE (& SCSI) (to avoid 2 drivers twiddling w/same phys backend... not good!).
>

Who is requiring that the boot device is spec'd as hda and not xvda?
I don't think there is such limitation in xend or libxl at the moment.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Don Dutile on
Stefano Stabellini wrote:
> On Thu, 1 Jul 2010, Don Dutile wrote:
>> The problem with this check/enable is that if you run
>> this on an older qemu-xen that doesn't have unplug support,
>> it fails pv-hvm configuration.
>>
>> But, all that means is that you can't use an xvd as the boot device,
>> and you have to use an emulated IDE device as boot device.
>> There are a couple ways to configure the vnif correctly (in guest
>> or in xen guest config file).
>>
>> So, on an older (say, rhel5) xen, I don't have this check;
>> the boot device is required to be spec'd as hda, not vda, and
>> xen-blkfront is not allowed to configure blk major nums
>> for IDE (& SCSI) (to avoid 2 drivers twiddling w/same phys backend... not good!).
>>
>
> Who is requiring that the boot device is spec'd as hda and not xvda?
> I don't think there is such limitation in xend or libxl at the moment.
>

If you take a previous xen HVM guest spec & just run it on a guest
that has pv-hvm added to it, then one has hda spec'd as boot device (by default,
by not editing the guest config spec/file).

Ideally, both config specs should/would work.

- Don

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Stefano Stabellini on
On Wed, 7 Jul 2010, Don Dutile wrote:
> Stefano Stabellini wrote:
> > On Thu, 1 Jul 2010, Don Dutile wrote:
> >> The problem with this check/enable is that if you run
> >> this on an older qemu-xen that doesn't have unplug support,
> >> it fails pv-hvm configuration.
> >>
> >> But, all that means is that you can't use an xvd as the boot device,
> >> and you have to use an emulated IDE device as boot device.
> >> There are a couple ways to configure the vnif correctly (in guest
> >> or in xen guest config file).
> >>
> >> So, on an older (say, rhel5) xen, I don't have this check;
> >> the boot device is required to be spec'd as hda, not vda, and
> >> xen-blkfront is not allowed to configure blk major nums
> >> for IDE (& SCSI) (to avoid 2 drivers twiddling w/same phys backend... not good!).
> >>
> >
> > Who is requiring that the boot device is spec'd as hda and not xvda?
> > I don't think there is such limitation in xend or libxl at the moment.
> >
>
> If you take a previous xen HVM guest spec & just run it on a guest
> that has pv-hvm added to it, then one has hda spec'd as boot device (by default,
> by not editing the guest config spec/file).
>
> Ideally, both config specs should/would work.

I wouldn't want to cause data corruptions by default to people that
specified xvda in their HVM config files by mistake (for example because
they copied and pasted from a PV guest config file).
But I agree that we should be able to do the right thing in your case
scenario too.

I propose the appended patch (to be merge with "Unplug emulated disks
and nics"): if the user specifies xen_emul_unplug=ignore and the unplug
protocol is not supported (old xen installations like rhel5), we
continue with the PV on HVM initialization and we make sure that
blkfront doesn't hook any IDE or SCSI device.

Don, Jeremy, what do you think about it?


---


diff --git a/arch/x86/xen/platform-pci-unplug.c b/arch/x86/xen/platform-pci-unplug.c
index 72a3da6..2f7f3fb 100644
--- a/arch/x86/xen/platform-pci-unplug.c
+++ b/arch/x86/xen/platform-pci-unplug.c
@@ -29,9 +29,9 @@
#define XEN_PLATFORM_ERR_PROTOCOL -2
#define XEN_PLATFORM_ERR_BLACKLIST -3

-/* boolean to signal that the platform pci device can be used */
-bool xen_platform_pci_enabled;
-EXPORT_SYMBOL_GPL(xen_platform_pci_enabled);
+/* store the value of xen_emul_unplug after the unplug is done */
+int xen_platform_pci_unplug;
+EXPORT_SYMBOL_GPL(xen_platform_pci_unplug);
static int xen_emul_unplug;

static int __init check_platform_magic(void)
@@ -76,13 +76,13 @@ void __init xen_unplug_emulated_devices(void)
/* If the version matches enable the Xen platform PCI driver.
* Also enable the Xen platform PCI driver if the version is really old
* and the user told us to ignore it. */
- if (!r || (r == XEN_PLATFORM_ERR_MAGIC &&
- (xen_emul_unplug & XEN_UNPLUG_IGNORE)))
- xen_platform_pci_enabled = 1;
+ if (r && !(r == XEN_PLATFORM_ERR_MAGIC &&
+ (xen_emul_unplug & XEN_UNPLUG_IGNORE)))
+ return;
/* Set the default value of xen_emul_unplug depending on whether or
* not the Xen PV frontends and the Xen platform PCI driver have
* been compiled for this kernel (modules or built-in are both OK). */
- if (xen_platform_pci_enabled && !xen_emul_unplug) {
+ if (!xen_emul_unplug) {
if (xen_must_unplug_nics()) {
printk(KERN_INFO "Netfront and the Xen platform PCI driver have "
"been compiled for this kernel: unplug emulated NICs.\n");
@@ -98,8 +98,9 @@ void __init xen_unplug_emulated_devices(void)
}
}
/* Now unplug the emulated devices */
- if (xen_platform_pci_enabled && !(xen_emul_unplug & XEN_UNPLUG_IGNORE))
+ if (!(xen_emul_unplug & XEN_UNPLUG_IGNORE))
outw(xen_emul_unplug, XEN_IOPORT_UNPLUG);
+ xen_platform_pci_unplug = xen_emul_unplug;
}

static int __init parse_xen_emul_unplug(char *arg)
diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c
index 82ed403..6eb2989 100644
--- a/drivers/block/xen-blkfront.c
+++ b/drivers/block/xen-blkfront.c
@@ -48,6 +48,7 @@
#include <xen/grant_table.h>
#include <xen/events.h>
#include <xen/page.h>
+#include <xen/platform_pci.h>

#include <xen/interface/grant_table.h>
#include <xen/interface/io/blkif.h>
@@ -737,6 +738,22 @@ static int blkfront_probe(struct xenbus_device *dev,
}
}

+ /* no unplug has been done: do not hook devices != xen vbds */
+ if (xen_hvm_domain() && (xen_platform_pci_unplug & XEN_UNPLUG_IGNORE)) {
+ int major;
+
+ if (!VDEV_IS_EXTENDED(vdevice))
+ major = BLKIF_MAJOR(vdevice);
+ else
+ major = XENVBD_MAJOR;
+
+ if (major != XENVBD_MAJOR) {
+ printk(KERN_INFO
+ "%s: HVM does not support vbd %d as xen block device\n",
+ __FUNCTION__, vdevice);
+ return -ENODEV;
+ }
+ }
info = kzalloc(sizeof(*info), GFP_KERNEL);
if (!info) {
xenbus_dev_fatal(dev, -ENOMEM, "allocating info structure");
diff --git a/drivers/xen/platform-pci.c b/drivers/xen/platform-pci.c
index be8b4f3..c01b5dd 100644
--- a/drivers/xen/platform-pci.c
+++ b/drivers/xen/platform-pci.c
@@ -196,7 +196,9 @@ static struct pci_driver platform_driver = {

static int __init platform_pci_module_init(void)
{
- if (!xen_platform_pci_enabled)
+ /* no unplug has been done, IGNORE hasn't been specified: just
+ * return now */
+ if (!xen_platform_pci_unplug)
return -ENODEV;

return pci_register_driver(&platform_driver);
diff --git a/drivers/xen/xenbus/xenbus_probe.c b/drivers/xen/xenbus/xenbus_probe.c
index 243279a..34287da 100644
--- a/drivers/xen/xenbus/xenbus_probe.c
+++ b/drivers/xen/xenbus/xenbus_probe.c
@@ -978,7 +978,7 @@ static void wait_for_devices(struct xenbus_driver *xendrv)
#ifndef MODULE
static int __init boot_wait_for_devices(void)
{
- if (xen_hvm_domain() && !xen_platform_pci_enabled)
+ if (xen_hvm_domain() && !xen_platform_pci_unplug)
return -ENODEV;

ready_to_wait_for_devices = 1;
diff --git a/include/xen/platform_pci.h b/include/xen/platform_pci.h
index afa8855..ce9d671 100644
--- a/include/xen/platform_pci.h
+++ b/include/xen/platform_pci.h
@@ -44,6 +44,6 @@ static inline int xen_must_unplug_disks(void) {
#endif
}

-extern bool xen_platform_pci_enabled;
+extern int xen_platform_pci_unplug;

#endif /* _XEN_PLATFORM_PCI_H */
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Don Dutile on
Stefano Stabellini wrote:
> On Wed, 7 Jul 2010, Don Dutile wrote:
>> Stefano Stabellini wrote:
>>> On Thu, 1 Jul 2010, Don Dutile wrote:
>>>> The problem with this check/enable is that if you run
>>>> this on an older qemu-xen that doesn't have unplug support,
>>>> it fails pv-hvm configuration.
>>>>
>>>> But, all that means is that you can't use an xvd as the boot device,
>>>> and you have to use an emulated IDE device as boot device.
>>>> There are a couple ways to configure the vnif correctly (in guest
>>>> or in xen guest config file).
>>>>
>>>> So, on an older (say, rhel5) xen, I don't have this check;
>>>> the boot device is required to be spec'd as hda, not vda, and
>>>> xen-blkfront is not allowed to configure blk major nums
>>>> for IDE (& SCSI) (to avoid 2 drivers twiddling w/same phys backend... not good!).
>>>>
>>> Who is requiring that the boot device is spec'd as hda and not xvda?
>>> I don't think there is such limitation in xend or libxl at the moment.
>>>
>> If you take a previous xen HVM guest spec & just run it on a guest
>> that has pv-hvm added to it, then one has hda spec'd as boot device (by default,
>> by not editing the guest config spec/file).
>>
>> Ideally, both config specs should/would work.
>
> I wouldn't want to cause data corruptions by default to people that
> specified xvda in their HVM config files by mistake (for example because
> they copied and pasted from a PV guest config file).
> But I agree that we should be able to do the right thing in your case
> scenario too.
>
> I propose the appended patch (to be merge with "Unplug emulated disks
> and nics"): if the user specifies xen_emul_unplug=ignore and the unplug
> protocol is not supported (old xen installations like rhel5), we
> continue with the PV on HVM initialization and we make sure that
> blkfront doesn't hook any IDE or SCSI device.
>
> Don, Jeremy, what do you think about it?
>
>

I guess what I'm wondering is why not set xen_emul_unplug to ignore by
default (static int xen_emul_unplug=XEN_UNPLUG_IGNORE), which handles
the case I mentioned (just take existing guest config file as is, no edits,
pre-pv-hvm added to guest kernel), and if person edits config file to
change boot device to xvda, they would then edit the config to add
-x xen_emul_unplug=[all|ide-disks|...] as well.

overall, i like the direction of the change to the unplug patch;
there'd be minor changes if you adopted the above default, but effectively
what's below would work well on older (rhel5) dom0's/tools.

- Don

> ---
>
>
> diff --git a/arch/x86/xen/platform-pci-unplug.c b/arch/x86/xen/platform-pci-unplug.c
> index 72a3da6..2f7f3fb 100644
> --- a/arch/x86/xen/platform-pci-unplug.c
> +++ b/arch/x86/xen/platform-pci-unplug.c
> @@ -29,9 +29,9 @@
> #define XEN_PLATFORM_ERR_PROTOCOL -2
> #define XEN_PLATFORM_ERR_BLACKLIST -3
>
> -/* boolean to signal that the platform pci device can be used */
> -bool xen_platform_pci_enabled;
> -EXPORT_SYMBOL_GPL(xen_platform_pci_enabled);
> +/* store the value of xen_emul_unplug after the unplug is done */
> +int xen_platform_pci_unplug;
> +EXPORT_SYMBOL_GPL(xen_platform_pci_unplug);
> static int xen_emul_unplug;
>
> static int __init check_platform_magic(void)
> @@ -76,13 +76,13 @@ void __init xen_unplug_emulated_devices(void)
> /* If the version matches enable the Xen platform PCI driver.
> * Also enable the Xen platform PCI driver if the version is really old
> * and the user told us to ignore it. */
> - if (!r || (r == XEN_PLATFORM_ERR_MAGIC &&
> - (xen_emul_unplug & XEN_UNPLUG_IGNORE)))
> - xen_platform_pci_enabled = 1;
> + if (r && !(r == XEN_PLATFORM_ERR_MAGIC &&
> + (xen_emul_unplug & XEN_UNPLUG_IGNORE)))
> + return;

and it should return ??? ^^^^


> /* Set the default value of xen_emul_unplug depending on whether or
> * not the Xen PV frontends and the Xen platform PCI driver have
> * been compiled for this kernel (modules or built-in are both OK). */
> - if (xen_platform_pci_enabled && !xen_emul_unplug) {
> + if (!xen_emul_unplug) {
> if (xen_must_unplug_nics()) {
> printk(KERN_INFO "Netfront and the Xen platform PCI driver have "
> "been compiled for this kernel: unplug emulated NICs.\n");
> @@ -98,8 +98,9 @@ void __init xen_unplug_emulated_devices(void)
> }
> }
> /* Now unplug the emulated devices */
> - if (xen_platform_pci_enabled && !(xen_emul_unplug & XEN_UNPLUG_IGNORE))
> + if (!(xen_emul_unplug & XEN_UNPLUG_IGNORE))
> outw(xen_emul_unplug, XEN_IOPORT_UNPLUG);
> + xen_platform_pci_unplug = xen_emul_unplug;
> }
>
> static int __init parse_xen_emul_unplug(char *arg)
> diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c
> index 82ed403..6eb2989 100644
> --- a/drivers/block/xen-blkfront.c
> +++ b/drivers/block/xen-blkfront.c
> @@ -48,6 +48,7 @@
> #include <xen/grant_table.h>
> #include <xen/events.h>
> #include <xen/page.h>
> +#include <xen/platform_pci.h>
>
> #include <xen/interface/grant_table.h>
> #include <xen/interface/io/blkif.h>
> @@ -737,6 +738,22 @@ static int blkfront_probe(struct xenbus_device *dev,
> }
> }
>
> + /* no unplug has been done: do not hook devices != xen vbds */
> + if (xen_hvm_domain() && (xen_platform_pci_unplug & XEN_UNPLUG_IGNORE)) {
> + int major;
> +
> + if (!VDEV_IS_EXTENDED(vdevice))
> + major = BLKIF_MAJOR(vdevice);
> + else
> + major = XENVBD_MAJOR;
> +
> + if (major != XENVBD_MAJOR) {
> + printk(KERN_INFO
> + "%s: HVM does not support vbd %d as xen block device\n",
> + __FUNCTION__, vdevice);
> + return -ENODEV;
> + }
> + }

Close to what I have for rhel5 (except I specifically scanned for ide & scsi blk major);
I like the above better -- xen-blkfront only supports.... xvd's ! :)


> info = kzalloc(sizeof(*info), GFP_KERNEL);
> if (!info) {
> xenbus_dev_fatal(dev, -ENOMEM, "allocating info structure");
> diff --git a/drivers/xen/platform-pci.c b/drivers/xen/platform-pci.c
> index be8b4f3..c01b5dd 100644
> --- a/drivers/xen/platform-pci.c
> +++ b/drivers/xen/platform-pci.c
> @@ -196,7 +196,9 @@ static struct pci_driver platform_driver = {
>
> static int __init platform_pci_module_init(void)
> {
> - if (!xen_platform_pci_enabled)
> + /* no unplug has been done, IGNORE hasn't been specified: just
> + * return now */
> + if (!xen_platform_pci_unplug)
> return -ENODEV;
>
> return pci_register_driver(&platform_driver);
> diff --git a/drivers/xen/xenbus/xenbus_probe.c b/drivers/xen/xenbus/xenbus_probe.c
> index 243279a..34287da 100644
> --- a/drivers/xen/xenbus/xenbus_probe.c
> +++ b/drivers/xen/xenbus/xenbus_probe.c
> @@ -978,7 +978,7 @@ static void wait_for_devices(struct xenbus_driver *xendrv)
> #ifndef MODULE
> static int __init boot_wait_for_devices(void)
> {
> - if (xen_hvm_domain() && !xen_platform_pci_enabled)
> + if (xen_hvm_domain() && !xen_platform_pci_unplug)
> return -ENODEV;
>
> ready_to_wait_for_devices = 1;
> diff --git a/include/xen/platform_pci.h b/include/xen/platform_pci.h
> index afa8855..ce9d671 100644
> --- a/include/xen/platform_pci.h
> +++ b/include/xen/platform_pci.h
> @@ -44,6 +44,6 @@ static inline int xen_must_unplug_disks(void) {
> #endif
> }
>
> -extern bool xen_platform_pci_enabled;
> +extern int xen_platform_pci_unplug;
>
> #endif /* _XEN_PLATFORM_PCI_H */
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel(a)lists.xensource.com
> http://lists.xensource.com/xen-devel

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/