From: Pranjal Shrivastava <praan@google.com>
To: Robin Murphy <robin.murphy@arm.com>
Cc: Joerg Roedel <joro@8bytes.org>, Will Deacon <will@kernel.org>,
Mostafa Saleh <smostafa@google.com>,
Nicolin Chen <nicolinc@nvidia.com>,
iommu@lists.linux.dev, Jason Gunthorpe <jgg@nvidia.com>,
Daniel Mentz <danielmentz@google.com>
Subject: Re: [PATCH v4 1/3] iommu/arm-smmu-v3: Introduce struct arm_smmu_event
Date: Fri, 1 Nov 2024 15:08:26 +0000 [thread overview]
Message-ID: <ZyTu6tNO8gaSqKSy@google.com> (raw)
In-Reply-To: <7c7b905c-ee00-4333-987e-0d0f0e45753f@arm.com>
On Fri, Nov 01, 2024 at 02:41:07PM +0000, Robin Murphy wrote:
> On 2024-10-18 7:00 pm, Pranjal Shrivastava wrote:
> > Introduce `struct arm_smmu_event` to represent event records.
> > Parse out relevant fields from raw event records for ease and
> > use the new `struct arm_smmu_event` instead.
> >
> > Signed-off-by: Daniel Mentz <danielmentz@google.com>
> > Signed-off-by: Pranjal Shrivastava <praan@google.com>
> > ---
> > drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 59 ++++++++++++++-------
> > drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h | 20 +++++++
> > 2 files changed, 59 insertions(+), 20 deletions(-)
> >
> > diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
> > index 737c5b882355..2f1108e5de51 100644
> > --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
> > +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
> > @@ -1757,17 +1757,16 @@ arm_smmu_find_master(struct arm_smmu_device *smmu, u32 sid)
> > }
> > /* IRQ and event handlers */
> > -static int arm_smmu_handle_evt(struct arm_smmu_device *smmu, u64 *evt)
> > +static int arm_smmu_handle_evt(struct arm_smmu_event *event)
> > {
> > int ret = 0;
> > u32 perm = 0;
> > struct arm_smmu_master *master;
> > - bool ssid_valid = evt[0] & EVTQ_0_SSV;
> > - u32 sid = FIELD_GET(EVTQ_0_SID, evt[0]);
> > struct iopf_fault fault_evt = { };
> > + struct arm_smmu_device *smmu = event->smmu;
> > struct iommu_fault *flt = &fault_evt.fault;
> > - switch (FIELD_GET(EVTQ_0_ID, evt[0])) {
> > + switch (event->id) {
> > case EVT_ID_TRANSLATION_FAULT:
> > case EVT_ID_ADDR_SIZE_FAULT:
> > case EVT_ID_ACCESS_FAULT:
> > @@ -1777,35 +1776,35 @@ static int arm_smmu_handle_evt(struct arm_smmu_device *smmu, u64 *evt)
> > return -EOPNOTSUPP;
> > }
> > - if (!(evt[1] & EVTQ_1_STALL))
> > + if (!event->stall)
> > return -EOPNOTSUPP;
> > - if (evt[1] & EVTQ_1_RnW)
> > + if (event->read)
> > perm |= IOMMU_FAULT_PERM_READ;
> > else
> > perm |= IOMMU_FAULT_PERM_WRITE;
> > - if (evt[1] & EVTQ_1_InD)
> > + if (event->instruction)
> > perm |= IOMMU_FAULT_PERM_EXEC;
> > - if (evt[1] & EVTQ_1_PnU)
> > + if (event->privileged)
> > perm |= IOMMU_FAULT_PERM_PRIV;
> > flt->type = IOMMU_FAULT_PAGE_REQ;
> > flt->prm = (struct iommu_fault_page_request) {
> > .flags = IOMMU_FAULT_PAGE_REQUEST_LAST_PAGE,
> > - .grpid = FIELD_GET(EVTQ_1_STAG, evt[1]),
> > + .grpid = event->stag,
> > .perm = perm,
> > - .addr = FIELD_GET(EVTQ_2_ADDR, evt[2]),
> > + .addr = event->iova,
> > };
> > - if (ssid_valid) {
> > + if (event->ssid_valid) {
> > flt->prm.flags |= IOMMU_FAULT_PAGE_REQUEST_PASID_VALID;
> > - flt->prm.pasid = FIELD_GET(EVTQ_0_SSID, evt[0]);
> > + flt->prm.pasid = event->ssid;
> > }
> > mutex_lock(&smmu->streams_mutex);
> > - master = arm_smmu_find_master(smmu, sid);
> > + master = arm_smmu_find_master(smmu, event->sid);
> > if (!master) {
> > ret = -EINVAL;
> > goto out_unlock;
> > @@ -1817,28 +1816,48 @@ static int arm_smmu_handle_evt(struct arm_smmu_device *smmu, u64 *evt)
> > return ret;
> > }
> > +static void arm_smmu_get_event_from_raw(struct arm_smmu_device *smmu,
> > + struct arm_smmu_event *event)
>
> One would kind of expect "get event from raw" to take a raw thing and return
> an event... personally I'd still just inline this in arm_smmu_handle_evt()
Hmm.. I think we can do both of those things.
We can kzalloc struct arm_smmu_event and return it, that should also
reduce the stack size. However, I'm unsure if that'd slow things down
because kzalloc may be a little more expensive than a local variable..
For the inlining, just to ensure I understand, we're looking to keep the
old arm_smmu_handle_evt(smmu, raw_evt) as is and then decode the event
within arm_smmu_handle_evt before we start handling it, right?
I'm fine with both of the suggestions above, let me know your vote?
> itself, but otherwise something like arm_smmu_decode_evt() would seem a more
> logical and obvious name at this point.
>
> > +{
> > + /* Pick out the good stuff */
> > + event->id = FIELD_GET(EVTQ_0_ID, event->raw[0]);
> > + event->sid = FIELD_GET(EVTQ_0_SID, event->raw[0]);
> > + event->ssid_valid = event->raw[0] & EVTQ_0_SSV;
> > + event->ssid = event->ssid_valid ? FIELD_GET(EVTQ_0_SSID, event->raw[0]) : IOMMU_NO_PASID;
> > + event->privileged = FIELD_GET(EVTQ_1_PnU, event->raw[1]);
> > + event->instruction = FIELD_GET(EVTQ_1_InD, event->raw[1]);
> > + event->s2 = FIELD_GET(EVTQ_1_S2, event->raw[1]);
> > + event->read = FIELD_GET(EVTQ_1_RnW, event->raw[1]);
> > + event->stag = FIELD_GET(EVTQ_1_STAG, event->raw[1]);
> > + event->stall = event->raw[1] & EVTQ_1_STALL;
> > + event->class = FIELD_GET(EVTQ_1_CLASS, event->raw[1]);
> > + event->iova = FIELD_GET(EVTQ_2_ADDR, event->raw[2]);
> > + event->ipa = FIELD_GET(EVTQ_3_IPA, event->raw[3]);
> > + event->smmu = smmu;
>
> The SMMU pointer isn't part of the raw event record... TBH I'd be inclined
> to leave it entirely separate, but if you really do want to hide it in the
> arm_smmu_event, at least keep things simple and initialise it outside the
> loop - it's not like it's ever going to change between events.
Yea.. honestly, just init-ing one member at a different time makes it
look a little weird if we allocate `arm_smmu_event` dynamically.
If we go ahead with allocating the `arm_smmu_event` using k*alloc, I
think it's best to remove the `smmu` member from the struct and pass it
around in functions.
>
> Thanks,
> Robin.
>
Thanks,
Praan
> > +}
> > +
> > static irqreturn_t arm_smmu_evtq_thread(int irq, void *dev)
> > {
> > int i, ret;
> > + struct arm_smmu_event evt;
> > struct arm_smmu_device *smmu = dev;
> > struct arm_smmu_queue *q = &smmu->evtq.q;
> > struct arm_smmu_ll_queue *llq = &q->llq;
> > static DEFINE_RATELIMIT_STATE(rs, DEFAULT_RATELIMIT_INTERVAL,
> > DEFAULT_RATELIMIT_BURST);
> > - u64 evt[EVTQ_ENT_DWORDS];
> > do {
> > - while (!queue_remove_raw(q, evt)) {
> > - u8 id = FIELD_GET(EVTQ_0_ID, evt[0]);
> > + while (!queue_remove_raw(q, evt.raw)) {
> > - ret = arm_smmu_handle_evt(smmu, evt);
> > + arm_smmu_get_event_from_raw(smmu, &evt);
> > + ret = arm_smmu_handle_evt(&evt);
> > if (!ret || !__ratelimit(&rs))
> > continue;
> > - dev_info(smmu->dev, "event 0x%02x received:\n", id);
> > - for (i = 0; i < ARRAY_SIZE(evt); ++i)
> > + dev_info(smmu->dev, "event 0x%02x received:\n", evt.id);
> > + for (i = 0; i < EVTQ_ENT_DWORDS; ++i)
> > dev_info(smmu->dev, "\t0x%016llx\n",
> > - (unsigned long long)evt[i]);
> > + (unsigned long long)evt.raw[i]);
> > cond_resched();
> > }
> > diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h
> > index 1e9952ca989f..8a42d7b701fb 100644
> > --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h
> > +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h
> > @@ -437,6 +437,7 @@ static inline unsigned int arm_smmu_cdtab_l2_idx(unsigned int ssid)
> > #define EVTQ_0_ID GENMASK_ULL(7, 0)
> > +/* Events */
> > #define EVT_ID_TRANSLATION_FAULT 0x10
> > #define EVT_ID_ADDR_SIZE_FAULT 0x11
> > #define EVT_ID_ACCESS_FAULT 0x12
> > @@ -452,6 +453,7 @@ static inline unsigned int arm_smmu_cdtab_l2_idx(unsigned int ssid)
> > #define EVTQ_1_RnW (1UL << 35)
> > #define EVTQ_1_S2 (1UL << 39)
> > #define EVTQ_1_CLASS GENMASK_ULL(41, 40)
> > +#define EVTQ_1_CLASS_TT 0x1
> > #define EVTQ_1_TT_READ (1UL << 44)
> > #define EVTQ_2_ADDR GENMASK_ULL(63, 0)
> > #define EVTQ_3_IPA GENMASK_ULL(51, 12)
> > @@ -771,6 +773,24 @@ struct arm_smmu_stream {
> > struct rb_node node;
> > };
> > +struct arm_smmu_event {
> > + struct arm_smmu_device *smmu;
> > + u8 id;
> > + u8 class;
> > + u16 stag;
> > + u32 sid;
> > + u32 ssid;
> > + u64 iova;
> > + u64 ipa;
> > + u64 raw[EVTQ_ENT_DWORDS];
> > + bool stall;
> > + bool ssid_valid;
> > + bool privileged;
> > + bool instruction;
> > + bool s2;
> > + bool read;
> > +};
> > +
> > /* SMMU private data for each master */
> > struct arm_smmu_master {
> > struct arm_smmu_device *smmu;
>
next prev parent reply other threads:[~2024-11-01 15:08 UTC|newest]
Thread overview: 47+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-10-18 18:00 [PATCH v4 0/3] iommu/arm-smmu-v3: Parse out event records Pranjal Shrivastava
2024-10-18 18:00 ` [PATCH v4 1/3] iommu/arm-smmu-v3: Introduce struct arm_smmu_event Pranjal Shrivastava
2024-10-19 1:56 ` Nicolin Chen
2024-10-21 6:20 ` Pranjal Shrivastava
2024-10-24 13:11 ` Will Deacon
2024-10-24 14:20 ` Pranjal Shrivastava
2024-10-24 17:02 ` Pranjal Shrivastava
2024-10-24 17:03 ` Jason Gunthorpe
2024-10-24 17:37 ` Pranjal Shrivastava
2024-10-28 12:23 ` Jason Gunthorpe
2024-10-28 14:46 ` Pranjal Shrivastava
2024-11-04 17:23 ` Daniel Mentz
2024-11-04 18:16 ` Pranjal Shrivastava
2024-11-04 18:19 ` Pranjal Shrivastava
2024-11-01 14:41 ` Robin Murphy
2024-11-01 15:08 ` Pranjal Shrivastava [this message]
2024-11-04 5:25 ` Daniel Mentz
2024-11-04 8:31 ` Pranjal Shrivastava
2024-11-07 0:10 ` Daniel Mentz
2024-11-07 14:33 ` Pranjal Shrivastava
2024-11-07 0:16 ` Daniel Mentz
2024-11-07 14:57 ` Pranjal Shrivastava
2024-11-11 22:20 ` Daniel Mentz
2024-11-12 0:52 ` Pranjal Shrivastava
2024-11-12 4:01 ` Daniel Mentz
2024-11-12 8:12 ` Pranjal Shrivastava
2024-10-18 18:00 ` [PATCH v4 2/3] iommu/arm-smmu-v3: Log better event records Pranjal Shrivastava
2024-10-19 2:06 ` Nicolin Chen
2024-10-19 4:51 ` Nicolin Chen
2024-10-21 6:29 ` Pranjal Shrivastava
2024-10-21 6:26 ` Pranjal Shrivastava
2024-10-21 22:53 ` Nicolin Chen
2024-10-24 13:15 ` Will Deacon
2024-10-24 14:14 ` Pranjal Shrivastava
2024-10-29 18:53 ` Will Deacon
2024-10-29 19:59 ` Pranjal Shrivastava
2024-10-24 19:00 ` Nicolin Chen
2024-10-29 18:49 ` Will Deacon
2024-11-01 15:05 ` Robin Murphy
2024-11-01 16:06 ` Pranjal Shrivastava
2024-11-04 6:36 ` Daniel Mentz
2024-11-04 10:51 ` Pranjal Shrivastava
2024-10-18 18:00 ` [PATCH v4 3/3] iommu/arm-smmu-v3: Avoid redundant master lookup in events Pranjal Shrivastava
2024-10-19 2:08 ` Nicolin Chen
2024-10-19 1:45 ` [PATCH v4 0/3] iommu/arm-smmu-v3: Parse out event records Nicolin Chen
2024-10-21 6:33 ` Pranjal Shrivastava
2024-10-21 22:51 ` Nicolin Chen
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=ZyTu6tNO8gaSqKSy@google.com \
--to=praan@google.com \
--cc=danielmentz@google.com \
--cc=iommu@lists.linux.dev \
--cc=jgg@nvidia.com \
--cc=joro@8bytes.org \
--cc=nicolinc@nvidia.com \
--cc=robin.murphy@arm.com \
--cc=smostafa@google.com \
--cc=will@kernel.org \
/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