qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Greg Kurz <groug@kaod.org>
To: David Gibson <david@gibson.dropbear.id.au>
Cc: qemu-devel@nongnu.org, qemu-ppc@nongnu.org,
	qemu-s390x@nongnu.org, "Alexey Kardashevskiy" <aik@ozlabs.ru>,
	"Cédric Le Goater" <clg@kaod.org>,
	"Michael Roth" <mdroth@linux.vnet.ibm.com>,
	"Paolo Bonzini" <pbonzini@redhat.com>,
	"Michael S. Tsirkin" <mst@redhat.com>,
	"Marcel Apfelbaum" <marcel@redhat.com>,
	"Eduardo Habkost" <ehabkost@redhat.com>,
	"David Hildenbrand" <david@redhat.com>,
	"Cornelia Huck" <cohuck@redhat.com>,
	"Gerd Hoffmann" <kraxel@redhat.com>,
	"Dmitry Fleytman" <dmitry.fleytman@gmail.com>,
	"Thomas Huth" <thuth@redhat.com>
Subject: Re: [Qemu-devel] [PATCH v3 03/19] xics: Disintricate allocation and type setting of interrupts
Date: Fri, 18 Jan 2019 12:47:47 +0100	[thread overview]
Message-ID: <20190118124747.2d78dfe6@bahia.lan> (raw)
In-Reply-To: <154774528595.1208625.15523908858345133758.stgit@bahia.lan>

On Thu, 17 Jan 2019 18:14:46 +0100
Greg Kurz <groug@kaod.org> wrote:

> The current code assumes that an interrupt is allocated as soon as its
> type is set to MSI or LSI. PHB hotplug will require to be able to set
> the type of an interrupt before actually allocating it.
> 
> Disintricate type setting from allocation by using another flag bit for
> the latter. Introduce a new ics_claim_irq() function for allocation. The
> behavior of aborting if the same irq gets allocated twice is kept.
> 
> ics_set_irq_type() now only sets the type to MSI or LSI. It doesn't bring
> anything to abort if the type was already set before. Drop the assert
> and XICS_FLAGS_IRQ_MASK on the way.
> 

With some more testing (and thinking :) I now realize that the type of
the irq should survive PHB unplug, which is not the case with this patch.

This calls for a v4.

> Older QEMUs don't know about XICS_FLAGS_IRQ_CLAIMED. In order to safely
> handle incoming migration, we must fix the irq flags. This is done at
> post load thanks to a compat property. We don't need to do anything for
> backward migration since older machines only care for the irq type.
> 
> Signed-off-by: Greg Kurz <groug@kaod.org>
> ---
>  hw/core/machine.c     |    4 ++++
>  hw/intc/xics.c        |   28 +++++++++++++++++++++++++---
>  hw/ppc/pnv_psi.c      |    1 +
>  hw/ppc/spapr_irq.c    |    4 +---
>  include/hw/ppc/xics.h |    8 ++++++--
>  5 files changed, 37 insertions(+), 8 deletions(-)
> 
> diff --git a/hw/core/machine.c b/hw/core/machine.c
> index 1a0a9ab1117a..536a34092367 100644
> --- a/hw/core/machine.c
> +++ b/hw/core/machine.c
> @@ -25,6 +25,10 @@
>  
>  GlobalProperty hw_compat_3_1[] = {
>      {
> +        .driver   = "ics-base",
> +        .property = "has-claimed-flag",
> +        .value    = "off",
> +    },{
>          .driver   = "pcie-root-port",
>          .property = "x-speed",
>          .value    = "2_5",
> diff --git a/hw/intc/xics.c b/hw/intc/xics.c
> index 16e8ffa2aaf7..82cf04548907 100644
> --- a/hw/intc/xics.c
> +++ b/hw/intc/xics.c
> @@ -75,7 +75,7 @@ void ics_pic_print_info(ICSState *ics, Monitor *mon)
>      for (i = 0; i < ics->nr_irqs; i++) {
>          ICSIRQState *irq = ics->irqs + i;
>  
> -        if (!(irq->flags & XICS_FLAGS_IRQ_MASK)) {
> +        if (!(irq->flags & XICS_FLAGS_IRQ_CLAIMED)) {
>              continue;
>          }
>          monitor_printf(mon, "  %4x %s %02x %02x\n",
> @@ -662,6 +662,22 @@ static int ics_base_dispatch_post_load(void *opaque, int version_id)
>      ICSState *ics = opaque;
>      ICSStateClass *info = ICS_BASE_GET_CLASS(ics);
>  
> +    if (!ics->has_claimed_flag) {
> +        int i;
> +
> +        for (i = 0; i < ics->nr_irqs; i++) {
> +            ICSIRQState *irq = ics->irqs + i;
> +
> +            /*
> +             * For older machines, allocating the irq and setting its type is
> +             * the same thing.
> +             */
> +            if (irq->flags & (XICS_FLAGS_IRQ_LSI| XICS_FLAGS_IRQ_MSI)) {
> +                irq->flags |= XICS_FLAGS_IRQ_CLAIMED;
> +            }
> +        }
> +    }
> +
>      if (info->post_load) {
>          return info->post_load(ics, version_id);
>      }
> @@ -702,6 +718,7 @@ static const VMStateDescription vmstate_ics_base = {
>  
>  static Property ics_base_properties[] = {
>      DEFINE_PROP_UINT32("nr-irqs", ICSState, nr_irqs, 0),
> +    DEFINE_PROP_BOOL("has-claimed-flag", ICSState, has_claimed_flag, true),
>      DEFINE_PROP_END_OF_LIST(),
>  };
>  
> @@ -741,10 +758,15 @@ ICPState *xics_icp_get(XICSFabric *xi, int server)
>      return xic->icp_get(xi, server);
>  }
>  
> -void ics_set_irq_type(ICSState *ics, int srcno, bool lsi)
> +void ics_claim_irq(ICSState *ics, int srcno)
>  {
> -    assert(!(ics->irqs[srcno].flags & XICS_FLAGS_IRQ_MASK));
> +    assert(ICS_IRQ_FREE(ics, srcno));
> +
> +    ics->irqs[srcno].flags |= XICS_FLAGS_IRQ_CLAIMED;
> +}
>  
> +void ics_set_irq_type(ICSState *ics, int srcno, bool lsi)
> +{
>      ics->irqs[srcno].flags |=
>          lsi ? XICS_FLAGS_IRQ_LSI : XICS_FLAGS_IRQ_MSI;
>  }
> diff --git a/hw/ppc/pnv_psi.c b/hw/ppc/pnv_psi.c
> index 8ced09506321..ced34e1119dc 100644
> --- a/hw/ppc/pnv_psi.c
> +++ b/hw/ppc/pnv_psi.c
> @@ -488,6 +488,7 @@ static void pnv_psi_realize(DeviceState *dev, Error **errp)
>  
>      for (i = 0; i < ics->nr_irqs; i++) {
>          ics_set_irq_type(ics, i, true);
> +        ics_claim_irq(ics, i);
>      }
>  
>      psi->qirqs = qemu_allocate_irqs(ics_simple_set_irq, ics, ics->nr_irqs);
> diff --git a/hw/ppc/spapr_irq.c b/hw/ppc/spapr_irq.c
> index 1da7a32348fc..86c712d15382 100644
> --- a/hw/ppc/spapr_irq.c
> +++ b/hw/ppc/spapr_irq.c
> @@ -125,9 +125,6 @@ error:
>      error_propagate(errp, local_err);
>  }
>  
> -#define ICS_IRQ_FREE(ics, srcno)   \
> -    (!((ics)->irqs[(srcno)].flags & (XICS_FLAGS_IRQ_MASK)))
> -
>  static int spapr_irq_claim_xics(sPAPRMachineState *spapr, int irq, bool lsi,
>                                  Error **errp)
>  {
> @@ -146,6 +143,7 @@ static int spapr_irq_claim_xics(sPAPRMachineState *spapr, int irq, bool lsi,
>      }
>  
>      ics_set_irq_type(ics, irq - ics->offset, lsi);
> +    ics_claim_irq(ics, irq - ics->offset);
>      return 0;
>  }
>  
> diff --git a/include/hw/ppc/xics.h b/include/hw/ppc/xics.h
> index fad786e8b22d..e3be5cc663c5 100644
> --- a/include/hw/ppc/xics.h
> +++ b/include/hw/ppc/xics.h
> @@ -131,6 +131,7 @@ struct ICSState {
>      /*< public >*/
>      uint32_t nr_irqs;
>      uint32_t offset;
> +    bool has_claimed_flag;
>      ICSIRQState *irqs;
>      XICSFabric *xics;
>  };
> @@ -153,13 +154,15 @@ struct ICSIRQState {
>  #define XICS_STATUS_PRESENTED          0x10
>  #define XICS_STATUS_QUEUED             0x20
>      uint8_t status;
> -/* (flags & XICS_FLAGS_IRQ_MASK) == 0 means the interrupt is not allocated */
>  #define XICS_FLAGS_IRQ_LSI             0x1
>  #define XICS_FLAGS_IRQ_MSI             0x2
> -#define XICS_FLAGS_IRQ_MASK            0x3
> +#define XICS_FLAGS_IRQ_CLAIMED         0x4
>      uint8_t flags;
>  };
>  
> +#define ICS_IRQ_FREE(ics, srcno)   \
> +    (!((ics)->irqs[(srcno)].flags & (XICS_FLAGS_IRQ_CLAIMED)))
> +
>  struct XICSFabric {
>      Object parent;
>  };
> @@ -193,6 +196,7 @@ void ics_simple_write_xive(ICSState *ics, int nr, int server,
>  void ics_simple_set_irq(void *opaque, int srcno, int val);
>  void ics_kvm_set_irq(void *opaque, int srcno, int val);
>  
> +void ics_claim_irq(ICSState *ics, int srcno);
>  void ics_set_irq_type(ICSState *ics, int srcno, bool lsi);
>  void icp_pic_print_info(ICPState *icp, Monitor *mon);
>  void ics_pic_print_info(ICSState *ics, Monitor *mon);
> 

  reply	other threads:[~2019-01-18 11:48 UTC|newest]

Thread overview: 47+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-01-17 17:14 [Qemu-devel] [PATCH v3 00/19] spapr: Add support for PHB hotplug Greg Kurz
2019-01-17 17:14 ` [Qemu-devel] [PATCH v3 01/19] ppc: Move spapr-related prototypes from xics.h into a seperate header file Greg Kurz
2019-01-17 17:14 ` [Qemu-devel] [PATCH v3 02/19] spapr: Rename xics to intc in interrupt controller agnostic code Greg Kurz
2019-01-18 13:56   ` Cédric Le Goater
2019-01-20 14:22     ` Greg Kurz
2019-02-05  6:03   ` David Gibson
2019-01-17 17:14 ` [Qemu-devel] [PATCH v3 03/19] xics: Disintricate allocation and type setting of interrupts Greg Kurz
2019-01-18 11:47   ` Greg Kurz [this message]
2019-01-18 12:26   ` Cédric Le Goater
2019-01-20 14:24     ` Greg Kurz
2019-02-05  6:13   ` David Gibson
2019-02-05 14:59     ` Greg Kurz
2019-02-06  1:47       ` David Gibson
2019-01-17 17:14 ` [Qemu-devel] [PATCH v3 04/19] spapr/xive: Don't set irq type in spapr_xive_irq_claim() Greg Kurz
2019-01-18 12:27   ` Cédric Le Goater
2019-01-17 17:14 ` [Qemu-devel] [PATCH v3 05/19] spapr: Set irq type in a dedicated function Greg Kurz
2019-01-18 12:34   ` Cédric Le Goater
2019-01-20 14:31     ` Greg Kurz
2019-01-17 17:15 ` [Qemu-devel] [PATCH v3 06/19] spapr: Identify LSIs of all possible PHBs at machine init Greg Kurz
2019-01-18 12:38   ` Cédric Le Goater
2019-01-20 14:37     ` Greg Kurz
2019-01-17 17:15 ` [Qemu-devel] [PATCH v3 07/19] spapr_pci: add PHB unrealize Greg Kurz
2019-01-18 11:54   ` Greg Kurz
2019-01-17 17:15 ` [Qemu-devel] [PATCH v3 08/19] spapr: create DR connectors for PHBs Greg Kurz
2019-01-17 17:15 ` [Qemu-devel] [PATCH v3 09/19] spapr: populate PHB DRC entries for root DT node Greg Kurz
2019-01-17 17:15 ` [Qemu-devel] [PATCH v3 10/19] spapr_events: add support for phb hotplug events Greg Kurz
2019-01-17 17:15 ` [Qemu-devel] [PATCH v3 11/19] qdev: pass an Object * to qbus_set_hotplug_handler() Greg Kurz
2019-01-17 17:15 ` [Qemu-devel] [PATCH v3 12/19] spapr_pci: provide node start offset via spapr_populate_pci_dt() Greg Kurz
2019-01-17 17:15 ` [Qemu-devel] [PATCH v3 13/19] spapr_pci: add ibm, my-drc-index property for PHB hotplug Greg Kurz
2019-01-17 17:15 ` [Qemu-devel] [PATCH v3 14/19] spapr: Factor out setting of "phandle" DT property to sPAPR irq frontend Greg Kurz
2019-01-18 12:45   ` Cédric Le Goater
2019-01-20 15:41     ` Greg Kurz
2019-01-17 17:16 ` [Qemu-devel] [PATCH v3 15/19] spapr_xive: Cache device tree nodename in sPAPRXive Greg Kurz
2019-01-18 13:38   ` Cédric Le Goater
2019-01-22 13:27     ` Greg Kurz
2019-01-22 14:26       ` Cédric Le Goater
2019-01-22 14:35         ` Greg Kurz
2019-01-17 17:16 ` [Qemu-devel] [PATCH v3 16/19] spapr: Expose the name of the interrupt controller node Greg Kurz
2019-01-18 13:44   ` Cédric Le Goater
2019-01-17 17:16 ` [Qemu-devel] [PATCH v3 17/19] spapr_irq: Expose the phandle of the interrupt controller Greg Kurz
2019-01-18 13:46   ` Cédric Le Goater
2019-01-22 13:32     ` Greg Kurz
2019-01-17 17:16 ` [Qemu-devel] [PATCH v3 18/19] spapr: add hotplug hooks for PHB hotplug Greg Kurz
2019-01-18 13:55   ` Cédric Le Goater
2019-01-17 17:16 ` [Qemu-devel] [PATCH v3 19/19] spapr: enable PHB hotplug for default pseries machine type Greg Kurz
2019-01-22  5:44 ` [Qemu-devel] [PATCH v3 00/19] spapr: Add support for PHB hotplug Alexey Kardashevskiy
2019-01-22  7:22   ` Greg Kurz

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20190118124747.2d78dfe6@bahia.lan \
    --to=groug@kaod.org \
    --cc=aik@ozlabs.ru \
    --cc=clg@kaod.org \
    --cc=cohuck@redhat.com \
    --cc=david@gibson.dropbear.id.au \
    --cc=david@redhat.com \
    --cc=dmitry.fleytman@gmail.com \
    --cc=ehabkost@redhat.com \
    --cc=kraxel@redhat.com \
    --cc=marcel@redhat.com \
    --cc=mdroth@linux.vnet.ibm.com \
    --cc=mst@redhat.com \
    --cc=pbonzini@redhat.com \
    --cc=qemu-devel@nongnu.org \
    --cc=qemu-ppc@nongnu.org \
    --cc=qemu-s390x@nongnu.org \
    --cc=thuth@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).