From: David Gibson <david@gibson.dropbear.id.au>
To: "Cédric Le Goater" <clg@kaod.org>
Cc: "Jason Wang" <jasowang@redhat.com>,
"Riku Voipio" <riku.voipio@iki.fi>,
qemu-devel@nongnu.org, groug@kaod.org, qemu-ppc@nongnu.org,
"Marc-André Lureau" <marcandre.lureau@redhat.com>,
"Paolo Bonzini" <pbonzini@redhat.com>,
philmd@redhat.com, "Laurent Vivier" <laurent@vivier.eu>
Subject: Re: [PATCH 09/20] spapr: Clarify and fix handling of nr_irqs
Date: Thu, 26 Sep 2019 11:03:24 +1000 [thread overview]
Message-ID: <20190926010324.GQ17405@umbus> (raw)
In-Reply-To: <55d983fb-3fd9-7442-cdd7-c2dcdb4a0ea9@kaod.org>
[-- Attachment #1: Type: text/plain, Size: 10272 bytes --]
On Wed, Sep 25, 2019 at 09:05:34AM +0200, Cédric Le Goater wrote:
> On 25/09/2019 08:45, David Gibson wrote:
> > Both the XICS and XIVE interrupt backends have a "nr-irqs" property, but
> > it means slightly different things. For XICS (or, strictly, the ICS) it
> > indicates the number of "real" external IRQs. Those start at XICS_IRQ_BASE
> > (0x1000) and don't include the special IPI vector. For XIVE, however, it
> > includes the whole IRQ space, including XIVE's many IPI vectors.
> >
> > The spapr code currently doesn't handle this sensibly, with the nr_irqs
> > value in SpaprIrq having different meanings depending on the backend.
> > We fix this by renaming nr_irqs to nr_xirqs and making it always indicate
> > just the number of external irqs, adjusting the value we pass to XIVE
> > accordingly. We also use move to using common constants in most of the
> > irq configurations, to make it clearer that the IRQ space looks the same
> > to the guest (and emulated devices), even if the backend is different.
> >
> > Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
>
> Reviewed-by: Cédric Le Goater <clg@kaod.org>
>
> one comment below,
>
> > ---
> > hw/ppc/spapr_irq.c | 48 +++++++++++++++-----------------------
> > include/hw/ppc/spapr_irq.h | 19 +++++++++------
> > 2 files changed, 31 insertions(+), 36 deletions(-)
> >
> > diff --git a/hw/ppc/spapr_irq.c b/hw/ppc/spapr_irq.c
> > index 8c26fa2d1e..5190a33e08 100644
> > --- a/hw/ppc/spapr_irq.c
> > +++ b/hw/ppc/spapr_irq.c
> > @@ -92,7 +92,7 @@ static void spapr_irq_init_kvm(SpaprMachineState *spapr,
> > * XICS IRQ backend.
> > */
> >
> > -static void spapr_irq_init_xics(SpaprMachineState *spapr, int nr_irqs,
> > +static void spapr_irq_init_xics(SpaprMachineState *spapr, int nr_xirqs,
> > Error **errp)
> > {
> > Object *obj;
> > @@ -102,7 +102,7 @@ static void spapr_irq_init_xics(SpaprMachineState *spapr, int nr_irqs,
> > object_property_add_child(OBJECT(spapr), "ics", obj, &error_abort);
> > object_property_add_const_link(obj, ICS_PROP_XICS, OBJECT(spapr),
> > &error_fatal);
> > - object_property_set_int(obj, nr_irqs, "nr-irqs", &error_fatal);
> > + object_property_set_int(obj, nr_xirqs, "nr-irqs", &error_fatal);
> > object_property_set_bool(obj, true, "realized", &local_err);
> > if (local_err) {
> > error_propagate(errp, local_err);
> > @@ -234,13 +234,9 @@ static void spapr_irq_init_kvm_xics(SpaprMachineState *spapr, Error **errp)
> > }
> > }
> >
> > -#define SPAPR_IRQ_XICS_NR_IRQS 0x1000
> > -#define SPAPR_IRQ_XICS_NR_MSIS \
> > - (XICS_IRQ_BASE + SPAPR_IRQ_XICS_NR_IRQS - SPAPR_IRQ_MSI)
> > -
> > SpaprIrq spapr_irq_xics = {
> > - .nr_irqs = SPAPR_IRQ_XICS_NR_IRQS,
> > - .nr_msis = SPAPR_IRQ_XICS_NR_MSIS,
> > + .nr_xirqs = SPAPR_NR_XIRQS,
> > + .nr_msis = SPAPR_NR_MSIS,
> > .ov5 = SPAPR_OV5_XIVE_LEGACY,
> >
> > .init = spapr_irq_init_xics,
> > @@ -260,7 +256,7 @@ SpaprIrq spapr_irq_xics = {
> > /*
> > * XIVE IRQ backend.
> > */
> > -static void spapr_irq_init_xive(SpaprMachineState *spapr, int nr_irqs,
> > +static void spapr_irq_init_xive(SpaprMachineState *spapr, int nr_xirqs,
> > Error **errp)
> > {
> > uint32_t nr_servers = spapr_max_server_number(spapr);
> > @@ -268,7 +264,7 @@ static void spapr_irq_init_xive(SpaprMachineState *spapr, int nr_irqs,
> > int i;
> >
> > dev = qdev_create(NULL, TYPE_SPAPR_XIVE);
> > - qdev_prop_set_uint32(dev, "nr-irqs", nr_irqs);
> > + qdev_prop_set_uint32(dev, "nr-irqs", nr_xirqs + SPAPR_XIRQ_BASE);
> > /*
> > * 8 XIVE END structures per CPU. One for each available priority
> > */
> > @@ -308,7 +304,7 @@ static qemu_irq spapr_qirq_xive(SpaprMachineState *spapr, int irq)
> > {
> > SpaprXive *xive = spapr->xive;
> >
> > - if (irq >= xive->nr_irqs) {
> > + if ((irq < SPAPR_XIRQ_BASE) || (irq >= xive->nr_irqs)) {
>
> So IPIs cannot be pulsed ? I think that is OK in QEMU.
They can be pulsed, they just can't be retrieved via the spapr_qirq()
interface. Since that interface basically exists for the spapr root
devices (VIO and PHBs) to find the qemu_irqs to wire themselves up to,
I think that's fine.
If we discover some reason we need to grab IPI qirqs by global number
then we can revisit this.
I'll add a comment to clarify this in the later patch where I unify
the qirq implementations.
> XIVE unifies all the interrupts at the controller level. Any one can trigger
> an interrupt with a store on the associate ESB page.
Absolutely, and nothing's stopping them.
>
> > return NULL;
> > }
> >
> > @@ -409,12 +405,9 @@ static void spapr_irq_init_kvm_xive(SpaprMachineState *spapr, Error **errp)
> > * with XICS.
> > */
> >
> > -#define SPAPR_IRQ_XIVE_NR_IRQS 0x2000
> > -#define SPAPR_IRQ_XIVE_NR_MSIS (SPAPR_IRQ_XIVE_NR_IRQS - SPAPR_IRQ_MSI)
> > -
> > SpaprIrq spapr_irq_xive = {
> > - .nr_irqs = SPAPR_IRQ_XIVE_NR_IRQS,
> > - .nr_msis = SPAPR_IRQ_XIVE_NR_MSIS,
> > + .nr_xirqs = SPAPR_NR_XIRQS,
> > + .nr_msis = SPAPR_NR_MSIS,
> > .ov5 = SPAPR_OV5_XIVE_EXPLOIT,
> >
> > .init = spapr_irq_init_xive,
> > @@ -450,18 +443,18 @@ static SpaprIrq *spapr_irq_current(SpaprMachineState *spapr)
> > &spapr_irq_xive : &spapr_irq_xics;
> > }
> >
> > -static void spapr_irq_init_dual(SpaprMachineState *spapr, int nr_irqs,
> > +static void spapr_irq_init_dual(SpaprMachineState *spapr, int nr_xirqs,
> > Error **errp)
> > {
> > Error *local_err = NULL;
> >
> > - spapr_irq_xics.init(spapr, spapr_irq_xics.nr_irqs, &local_err);
> > + spapr_irq_xics.init(spapr, spapr_irq_xics.nr_xirqs, &local_err);
> > if (local_err) {
> > error_propagate(errp, local_err);
> > return;
> > }
> >
> > - spapr_irq_xive.init(spapr, spapr_irq_xive.nr_irqs, &local_err);
> > + spapr_irq_xive.init(spapr, spapr_irq_xive.nr_xirqs, &local_err);
> > if (local_err) {
> > error_propagate(errp, local_err);
> > return;
> > @@ -586,12 +579,9 @@ static const char *spapr_irq_get_nodename_dual(SpaprMachineState *spapr)
> > /*
> > * Define values in sync with the XIVE and XICS backend
> > */
> > -#define SPAPR_IRQ_DUAL_NR_IRQS 0x2000
> > -#define SPAPR_IRQ_DUAL_NR_MSIS (SPAPR_IRQ_DUAL_NR_IRQS - SPAPR_IRQ_MSI)
> > -
> > SpaprIrq spapr_irq_dual = {
> > - .nr_irqs = SPAPR_IRQ_DUAL_NR_IRQS,
> > - .nr_msis = SPAPR_IRQ_DUAL_NR_MSIS,
> > + .nr_xirqs = SPAPR_NR_XIRQS,
> > + .nr_msis = SPAPR_NR_MSIS,
> > .ov5 = SPAPR_OV5_XIVE_BOTH,
> >
> > .init = spapr_irq_init_dual,
> > @@ -693,10 +683,10 @@ void spapr_irq_init(SpaprMachineState *spapr, Error **errp)
> > spapr_irq_msi_init(spapr, spapr->irq->nr_msis);
> > }
> >
> > - spapr->irq->init(spapr, spapr->irq->nr_irqs, errp);
> > + spapr->irq->init(spapr, spapr->irq->nr_xirqs, errp);
> >
> > spapr->qirqs = qemu_allocate_irqs(spapr->irq->set_irq, spapr,
> > - spapr->irq->nr_irqs);
> > + spapr->irq->nr_xirqs + SPAPR_XIRQ_BASE);
> > }
> >
> > int spapr_irq_claim(SpaprMachineState *spapr, int irq, bool lsi, Error **errp)
> > @@ -804,11 +794,11 @@ int spapr_irq_find(SpaprMachineState *spapr, int num, bool align, Error **errp)
> > return first + ics->offset;
> > }
> >
> > -#define SPAPR_IRQ_XICS_LEGACY_NR_IRQS 0x400
> > +#define SPAPR_IRQ_XICS_LEGACY_NR_XIRQS 0x400
> >
> > SpaprIrq spapr_irq_xics_legacy = {
> > - .nr_irqs = SPAPR_IRQ_XICS_LEGACY_NR_IRQS,
> > - .nr_msis = SPAPR_IRQ_XICS_LEGACY_NR_IRQS,
> > + .nr_xirqs = SPAPR_IRQ_XICS_LEGACY_NR_XIRQS,
> > + .nr_msis = SPAPR_IRQ_XICS_LEGACY_NR_XIRQS,
> > .ov5 = SPAPR_OV5_XIVE_LEGACY,
> >
> > .init = spapr_irq_init_xics,
> > diff --git a/include/hw/ppc/spapr_irq.h b/include/hw/ppc/spapr_irq.h
> > index 5db305165c..a8f9a2ab11 100644
> > --- a/include/hw/ppc/spapr_irq.h
> > +++ b/include/hw/ppc/spapr_irq.h
> > @@ -16,13 +16,18 @@
> > * IRQ range offsets per device type
> > */
> > #define SPAPR_IRQ_IPI 0x0
> > -#define SPAPR_IRQ_EPOW 0x1000 /* XICS_IRQ_BASE offset */
> > -#define SPAPR_IRQ_HOTPLUG 0x1001
> > -#define SPAPR_IRQ_VIO 0x1100 /* 256 VIO devices */
> > -#define SPAPR_IRQ_PCI_LSI 0x1200 /* 32+ PHBs devices */
> >
> > -#define SPAPR_IRQ_MSI 0x1300 /* Offset of the dynamic range covered
> > - * by the bitmap allocator */
> > +#define SPAPR_XIRQ_BASE XICS_IRQ_BASE /* 0x1000 */
> > +#define SPAPR_IRQ_EPOW (SPAPR_XIRQ_BASE + 0x0000)
> > +#define SPAPR_IRQ_HOTPLUG (SPAPR_XIRQ_BASE + 0x0001)
> > +#define SPAPR_IRQ_VIO (SPAPR_XIRQ_BASE + 0x0100) /* 256 VIO devices */
> > +#define SPAPR_IRQ_PCI_LSI (SPAPR_XIRQ_BASE + 0x0200) /* 32+ PHBs devices */
> > +
> > +/* Offset of the dynamic range covered by the bitmap allocator */
> > +#define SPAPR_IRQ_MSI (SPAPR_XIRQ_BASE + 0x0300)
> > +
> > +#define SPAPR_NR_XIRQS 0x1000
> > +#define SPAPR_NR_MSIS (SPAPR_XIRQ_BASE + SPAPR_NR_XIRQS - SPAPR_IRQ_MSI)
> >
> > typedef struct SpaprMachineState SpaprMachineState;
> >
> > @@ -32,7 +37,7 @@ int spapr_irq_msi_alloc(SpaprMachineState *spapr, uint32_t num, bool align,
> > void spapr_irq_msi_free(SpaprMachineState *spapr, int irq, uint32_t num);
> >
> > typedef struct SpaprIrq {
> > - uint32_t nr_irqs;
> > + uint32_t nr_xirqs;
> > uint32_t nr_msis;
> > uint8_t ov5;
> >
> >
>
--
David Gibson | I'll have my music baroque, and my code
david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_
| _way_ _around_!
http://www.ozlabs.org/~dgibson
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
next prev parent reply other threads:[~2019-09-26 1:31 UTC|newest]
Thread overview: 93+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-09-25 6:45 [PATCH 00/20] spapr: IRQ subsystem cleanups David Gibson
2019-09-25 6:45 ` [PATCH 01/20] xics: Use incomplete type for XICSFabric David Gibson
2019-09-25 6:55 ` Cédric Le Goater
2019-09-25 7:48 ` Greg Kurz
2019-09-25 7:45 ` Greg Kurz
2019-09-25 6:45 ` [PATCH 02/20] xics: Eliminate 'reject', 'resend' and 'eoi' class hooks David Gibson
2019-09-25 6:45 ` [PATCH 03/20] xics: Rename misleading ics_simple_*() functions David Gibson
2019-09-25 6:45 ` [PATCH 04/20] xics: Eliminate reset hook David Gibson
2019-09-25 7:33 ` Cédric Le Goater
2019-09-25 7:59 ` Greg Kurz
2019-09-26 2:54 ` David Gibson
2019-09-25 6:45 ` [PATCH 05/20] xics: Merge TYPE_ICS_BASE and TYPE_ICS_SIMPLE classes David Gibson
2019-09-25 8:16 ` Greg Kurz
2019-09-25 8:31 ` Greg Kurz
2019-09-26 0:55 ` David Gibson
2019-09-26 0:52 ` David Gibson
2019-09-25 12:47 ` Cédric Le Goater
2019-09-25 6:45 ` [PATCH 06/20] xics: Create sPAPR specific ICS subtype David Gibson
2019-09-25 7:34 ` Cédric Le Goater
2019-09-25 8:40 ` Greg Kurz
2019-09-25 8:55 ` Cédric Le Goater
2019-09-25 9:07 ` Greg Kurz
2019-09-26 0:56 ` David Gibson
2019-09-26 7:09 ` Cédric Le Goater
2019-09-27 16:05 ` Greg Kurz
2019-09-30 8:45 ` David Gibson
2019-09-30 17:00 ` Greg Kurz
2019-10-01 1:45 ` David Gibson
2019-09-25 6:45 ` [PATCH 07/20] spapr: Fold spapr_phb_lsi_qirq() into its single caller David Gibson
2019-09-25 6:58 ` Cédric Le Goater
2019-09-25 8:56 ` Greg Kurz
2019-09-26 7:08 ` Philippe Mathieu-Daudé
2019-09-25 6:45 ` [PATCH 08/20] spapr: Replace spapr_vio_qirq() helper with spapr_vio_irq_pulse() helper David Gibson
2019-09-25 6:58 ` Cédric Le Goater
2019-09-25 8:57 ` Greg Kurz
2019-09-26 7:08 ` Philippe Mathieu-Daudé
2019-09-25 6:45 ` [PATCH 09/20] spapr: Clarify and fix handling of nr_irqs David Gibson
2019-09-25 7:05 ` Cédric Le Goater
2019-09-26 1:03 ` David Gibson [this message]
2019-09-26 7:02 ` Cédric Le Goater
2019-09-25 17:13 ` Greg Kurz
2019-09-25 6:45 ` [PATCH 10/20] spapr: Eliminate nr_irqs parameter to SpaprIrq::init David Gibson
2019-09-25 7:06 ` Cédric Le Goater
2019-09-25 17:16 ` Greg Kurz
2019-09-25 6:45 ` [PATCH 11/20] spapr: Fix indexing of XICS irqs David Gibson
2019-09-25 7:11 ` Cédric Le Goater
2019-09-25 20:17 ` Greg Kurz
2019-09-26 1:31 ` David Gibson
2019-09-26 7:21 ` Greg Kurz
2019-09-26 11:32 ` David Gibson
2019-09-26 14:44 ` Greg Kurz
2019-09-25 6:45 ` [PATCH 12/20] spapr: Simplify spapr_qirq() handling David Gibson
2019-09-25 7:16 ` Cédric Le Goater
2019-09-25 20:30 ` Greg Kurz
2019-09-26 7:10 ` Philippe Mathieu-Daudé
2019-09-25 6:45 ` [PATCH 13/20] spapr: Eliminate SpaprIrq:get_nodename method David Gibson
2019-09-25 7:19 ` Cédric Le Goater
2019-09-26 7:11 ` Philippe Mathieu-Daudé
2019-09-26 7:48 ` Greg Kurz
2019-09-26 11:36 ` David Gibson
2019-09-25 6:45 ` [PATCH 14/20] spapr: Remove unhelpful tracepoints from spapr_irq_free_xics() David Gibson
2019-09-25 7:20 ` Cédric Le Goater
2019-09-26 7:11 ` Philippe Mathieu-Daudé
2019-09-26 7:50 ` Greg Kurz
2019-09-25 6:45 ` [PATCH 15/20] spapr: Handle freeing of multiple irqs in frontend only David Gibson
2019-09-25 7:21 ` Cédric Le Goater
2019-09-26 7:52 ` Greg Kurz
2019-09-25 6:45 ` [PATCH 16/20] spapr, xics, xive: Better use of assert()s on irq claim/free paths David Gibson
2019-09-25 7:22 ` Cédric Le Goater
2019-09-26 8:08 ` Greg Kurz
2019-09-26 11:39 ` David Gibson
2019-09-25 6:45 ` [PATCH 17/20] spapr: Remove unused return value in claim path David Gibson
2019-09-25 7:23 ` Cédric Le Goater
2019-09-26 7:13 ` Philippe Mathieu-Daudé
2019-09-26 8:36 ` Greg Kurz
2019-09-27 1:47 ` David Gibson
2019-09-25 6:45 ` [PATCH 18/20] xive: Improve irq claim/free path David Gibson
2019-09-25 7:25 ` Cédric Le Goater
2019-09-26 1:05 ` David Gibson
2019-09-25 6:45 ` [PATCH 19/20] spapr: Use less cryptic representation of which irq backends are supported David Gibson
2019-09-25 7:28 ` Cédric Le Goater
2019-09-26 9:16 ` Greg Kurz
2019-09-25 6:45 ` [PATCH 20/20] spapr: Eliminate SpaprIrq::init hook David Gibson
2019-09-25 7:31 ` Cédric Le Goater
2019-09-26 1:13 ` David Gibson
2019-09-26 7:05 ` Cédric Le Goater
2019-09-26 11:29 ` David Gibson
2019-09-26 15:35 ` Greg Kurz
2019-09-27 5:51 ` David Gibson
2019-09-27 6:23 ` Greg Kurz
2019-09-26 15:39 ` Greg Kurz
2019-09-27 14:12 ` Greg Kurz
2019-09-29 9:34 ` David Gibson
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=20190926010324.GQ17405@umbus \
--to=david@gibson.dropbear.id.au \
--cc=clg@kaod.org \
--cc=groug@kaod.org \
--cc=jasowang@redhat.com \
--cc=laurent@vivier.eu \
--cc=marcandre.lureau@redhat.com \
--cc=pbonzini@redhat.com \
--cc=philmd@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=qemu-ppc@nongnu.org \
--cc=riku.voipio@iki.fi \
/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).