Proposed patches for Linux 3.1 kernel.

Proposed patches for Linux 3.1 kernel.

[Konrad Rzeszutek Wilk]

The 3.0 is suppose to be released today, or perhaps tomorrow.

For 3.1 kernel we have this long list of features and bug-fixes that we are going
to ask Linus to pick up:

 - VGA text support. Since Xen is the first payload on the bootloader it gets the
   screen information - now with this patch we can copy it from the hypervisor
   and enjoy 80x25 old-school text output.
 - MODULE_ALIAS on backends so that udev will autoload them.
 - tmem driver to shuffle file-system and swap pages between guests as appropiate.
   Runtime it is disabled unless 'tmem' parameter is supplied.
 - self balloon driver to decrease memory in the guest and make the swap pages
   be shuffled by tmem to be compressed/shared/etc. Runtime wise it is disabled
   by default (use 'selfballooning' to enable it and make sure you have a swap disk).
 - pciback backend. It complements the xen pci front driver which together
   allow guests to have host's PCI devices.
 - cleaning up the Xen PCI glue code.
 - Fix section mismatch in Xen PCI HVM driver (xen-pci-platform).
 - Fixes to the generic pagetable to compute precisely how many pages we need.
 - blkback can now provide completion latency that follows the hardware's completion
   latency.
 - PVonHVM blkfront naming fixes.
 - Xen MMU debugfs has gotten a facelift - now it uses the tracing API.

Please tell me or Jeremy if we are missing anything or need to chase something down.

Here is the full credit list:

Anupam Chanda (1):
      x86, hyper: Change hypervisor detection order

Bastian Blank (4):
      xen: Populate xenbus device attributes
      xen: Add module alias to autoload backend drivers
      xen/blkback: Add module alias for autoloading
      xen/netback: Add module alias for autoloading

Dan Magenheimer (2):
      xen: prepare tmem shim to handle frontswap
      xen: tmem: self-ballooning and frontswap-selfshrinking

Daniel Stodden (1):
      xen/blkback: Don't let in-flight requests defer pending ones.

Ian Campbell (1):
      xen: rename pciback module to xen-pciback.

Jeremy Fitzhardinge (18):
      xen: allow enable use of VGA console on dom0
      trace/xen: add skeleton for Xen trace events
      xen/multicalls: remove debugfs stats
      xen/trace: set up tracepoint skeleton
      xen/trace: add multicall tracing
      xen/trace: add mmu tracepoints
      xen/trace: add ptpage alloc/release tracepoints
      xen/trace: add xen_pgd_(un)pin tracepoints
      xen/trace: add segment desc tracing
      xen/trace: add tlb flush tracepoints
      xen/mmu: use extend_args for more mmuext updates
      xen/mmu: tune pgtable alloc/release
      xen/multicalls: disable MC_DEBUG
      xen/multicalls: add unlikely around slowpath in __xen_mc_entry()
      xen/multicall: special-case singleton hypercalls
      xen/multicall: move *idx fields to start of mc_buffer
      xen/trace: convert mmu events to use DECLARE_EVENT_CLASS()/DEFINE_EVENT()
      xen/trace: use class for multicall trace

Joe Perches (1):
      xen: Add __attribute__((format(printf... where appropriate

Konrad Rzeszutek Wilk (21):
      xen/pci: Shuffle code around.
      xen/pci: Update comments and fix empty spaces.
      xen/pci: Provide #ifdef CONFIG_ACPI to easy code squashing.
      xen/pci: In xen_register_pirq bind the GSI to the IRQ after the hypercall.
      xen/pci: Use the xen_register_pirq for HVM and initial domain users
      xen/pci: Squash pci_xen_initial_domain and xen_setup_pirqs together.
      xen/pci: Move the allocation of IRQs when there are no IOAPIC's to the end
      xen/pci: Retire unnecessary #ifdef CONFIG_ACPI
      xen/pci: Remove 'xen_allocate_pirq_gsi'.
      xen/pci: Use 'acpi_gsi_to_irq' value unconditionally.
      xen/pciback: xen pci backend driver.
      xen/pciback: Cleanup the driver based on checkpatch warnings and errors.
      xen/pciback: Register the owner (domain) of the PCI device.
      xen/pciback: Disable MSI/MSI-X when reseting a device
      xen/pciback: Allocate IRQ handler for device that is shared with guest.
      xen/pciback: Fine-grain the spinlocks and fix BUG: scheduling while atomic cases.
      xen/pciback: Don't setup an fake IRQ handler for SR-IOV devices.
      xen/pciback: Print out the MSI/MSI-X (PIRQ) values
      xen/pciback: Drop two backends, squash and cleanup some code.
      xen/pciback: Remove the DEBUG option.
      xen/pciback: Have 'passthrough' option instead of XEN_PCIDEV_BACKEND_PASS and XEN_PCIDEV_BACKEND_VPCI

Raghavendra D Prabhu (1):
      xen:pvhvm: Modpost section mismatch fix

Stefan Bader (2):
      xen-blkfront: Drop name and minor adjustments for emulated scsi devices
      xen-blkfront: Fix one off warning about name clash

Stefano Stabellini (4):
      x86, mm: Calculate precisely the memory needed by init_memory_mapping
      Revert "x86,xen: introduce x86_init.mapping.pagetable_reserve"
      x86, init : Move memblock_x86_reserve_range PGTABLE to find_early_table_space
      x86-64, mm: Do not assume head_64.S used 4KB pages when !use_pse

Zhao, Yu (1):
      xen/pciback: guest SR-IOV support for PV guest

Re: [Xen-devel] Proposed patches for Linux 3.1 kernel.

[Sander Eikelenboom]

Thursday, July 21, 2011, 4:17:15 PM, you wrote:

> The 3.0 is suppose to be released today, or perhaps tomorrow.

> For 3.1 kernel we have this long list of features and bug-fixes that we are going
> to ask Linus to pick up:

>  - VGA text support. Since Xen is the first payload on the bootloader it gets the
>    screen information - now with this patch we can copy it from the hypervisor
>    and enjoy 80x25 old-school text output.

I assume this one will be proposed for 3.0.1 stable as well ?


>  - MODULE_ALIAS on backends so that udev will autoload them.
>  - tmem driver to shuffle file-system and swap pages between guests as appropiate.
>    Runtime it is disabled unless 'tmem' parameter is supplied.
>  - self balloon driver to decrease memory in the guest and make the swap pages
>    be shuffled by tmem to be compressed/shared/etc. Runtime wise it is disabled
>    by default (use 'selfballooning' to enable it and make sure you have a swap disk).
>  - pciback backend. It complements the xen pci front driver which together
>    allow guests to have host's PCI devices.
>  - cleaning up the Xen PCI glue code.
>  - Fix section mismatch in Xen PCI HVM driver (xen-pci-platform).
>  - Fixes to the generic pagetable to compute precisely how many pages we need.
>  - blkback can now provide completion latency that follows the hardware's completion
>    latency.
>  - PVonHVM blkfront naming fixes.
>  - Xen MMU debugfs has gotten a facelift - now it uses the tracing API.

> Please tell me or Jeremy if we are missing anything or need to chase something down.

Re: Proposed patches for Linux 3.1 kernel.

[Konrad Rzeszutek Wilk]

On Thu, Jul 21, 2011 at 04:43:59PM +0200, Sander Eikelenboom wrote:
> Thursday, July 21, 2011, 4:17:15 PM, you wrote:
> 
> > The 3.0 is suppose to be released today, or perhaps tomorrow.
> 
> > For 3.1 kernel we have this long list of features and bug-fixes that we are going
> > to ask Linus to pick up:
> 
> >  - VGA text support. Since Xen is the first payload on the bootloader it gets the
> >    screen information - now with this patch we can copy it from the hypervisor
> >    and enjoy 80x25 old-school text output.
> 
> I assume this one will be proposed for 3.0.1 stable as well ?

Sadly no - it is considered a feature - and those (as I understand, maybe I am
incorrect?) don't go on the stable wagon.

The vga patch is self-contained in here: http://darnok.org/xen/vga.patch if you would
like to patch your 3.0 kernel. Keep in mind - that you can still run stock 3.0 - with Xorg,
frame-buffer support (KMS), or with serial output.

Re: Proposed patches for Linux 3.1 kernel.

[Jeremy Fitzhardinge]

On 07/21/2011 10:59 AM, Konrad Rzeszutek Wilk wrote:
> On Thu, Jul 21, 2011 at 04:43:59PM +0200, Sander Eikelenboom wrote:
>> Thursday, July 21, 2011, 4:17:15 PM, you wrote:
>>
>>> The 3.0 is suppose to be released today, or perhaps tomorrow.
>>> For 3.1 kernel we have this long list of features and bug-fixes that we are going
>>> to ask Linus to pick up:
>>>  - VGA text support. Since Xen is the first payload on the bootloader it gets the
>>>    screen information - now with this patch we can copy it from the hypervisor
>>>    and enjoy 80x25 old-school text output.
>> I assume this one will be proposed for 3.0.1 stable as well ?
> Sadly no - it is considered a feature - and those (as I understand, maybe I am
> incorrect?) don't go on the stable wagon.

Eh, I think its probably fairly justifyable for stable, since its
completely self-contained and can't cause any regressions.  Its akin to
adding a new driver.  Wouldn't want to argue it for a late -rc, but
stable backport is probably OK.

> The vga patch is self-contained in here: http://darnok.org/xen/vga.patch if you would
> like to patch your 3.0 kernel. Keep in mind - that you can still run stock 3.0 - with Xorg,
> frame-buffer support (KMS), or with serial output.

What git branch is that on?

    J

Re: Proposed patches for Linux 3.1 kernel.

[Konrad Rzeszutek Wilk]

On Thu, Jul 21, 2011 at 12:04:16PM -0700, Jeremy Fitzhardinge wrote:
> On 07/21/2011 10:59 AM, Konrad Rzeszutek Wilk wrote:
> > On Thu, Jul 21, 2011 at 04:43:59PM +0200, Sander Eikelenboom wrote:
> >> Thursday, July 21, 2011, 4:17:15 PM, you wrote:
> >>
> >>> The 3.0 is suppose to be released today, or perhaps tomorrow.
> >>> For 3.1 kernel we have this long list of features and bug-fixes that we are going
> >>> to ask Linus to pick up:
> >>>  - VGA text support. Since Xen is the first payload on the bootloader it gets the
> >>>    screen information - now with this patch we can copy it from the hypervisor
> >>>    and enjoy 80x25 old-school text output.
> >> I assume this one will be proposed for 3.0.1 stable as well ?
> > Sadly no - it is considered a feature - and those (as I understand, maybe I am
> > incorrect?) don't go on the stable wagon.
> 
> Eh, I think its probably fairly justifyable for stable, since its
> completely self-contained and can't cause any regressions.  Its akin to
> adding a new driver.  Wouldn't want to argue it for a late -rc, but
> stable backport is probably OK.
> 
> > The vga patch is self-contained in here: http://darnok.org/xen/vga.patch if you would
> > like to patch your 3.0 kernel. Keep in mind - that you can still run stock 3.0 - with Xorg,
> > frame-buffer support (KMS), or with serial output.
> 
> What git branch is that on?

stable/vga.support.