Add PASID support to SMMUv3 unmanaged domains
Subsystems (3)
v4 cover Add PASID support to SMMUv3 unmanaged domains
Hi all,
This patch series implements the set_dev_pasid operation for DMA and UNMANAGED iommu domains.
The bulk of the series involves a refactor of stage 1 domains so that they describe a single CD entry. On attach, stage 1 domains are inserted into a CD table that is now owned by the arm_smmu_master struct. This is a pre-requisite to the set_dev_pasid implementation but also results in a conceptually cleaner arm_smmu_domain. Note that this does not preclude from attaching domains that represent a CD table, such as for the proposed iommufd NESTED domains.
The last few patches of the series make drive-by cleanups to the smmu SVA implementation. A follow-up patch-series is planned to further take advantage of these refactorings so that the SVA set_dev_pasid implementation can directly rely on the arm-smmu-v3.c's set_dev_pasid implementation. See discussion on patch 14 of the v2 series.
This patch series is also available on gerrit with Jean's SMMU test engine patches cherry-picked on top:
https://linux-review.googlesource.com/id/I0fcd9adc058d1c58a12d2599cc82fba73da7697a
This allowed testing of basic SVA functionality (e.g.: attaching, page fault handling, and detaching).
Thanks,
Michael Shavit
Changelog
v4:
* Fix build warning and error on patch 07. The error was introduced during a v1->v2 rebase and hidden by patch 09 which removed the offending line.
v3:
https://lore.kernel.org/all/[email protected]/
* Dropped the bulk of the SVA refactoring to re-work as a follow-up series.
* Reworded cover letter to omit dropped changes.
* Rebased on 6.4 tip
v2:
https://lore.kernel.org/all/[email protected]/
* Reworded cover letter and commits based on v1 feedback.
* Split and reworked iommu/arm-smmu-v3: Move cdtable to arm_smmu_master
* Added SVA clean-up and refactor.
* A few other small bug fixes and cosmetics.
v1:
https://lore.kernel.org/all/[email protected]/
Michael Shavit (13):
iommu/arm-smmu-v3: Move ctx_desc out of s1_cfg iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master iommu/arm-smmu-v3: Refactor write_strtab_ent iommu/arm-smmu-v3: Refactor write_ctx_desc iommu/arm-smmu-v3: Use the master-owned s1_cfg iommu/arm-smmu-v3: Simplify arm_smmu_enable_ats iommu/arm-smmu-v3: Keep track of attached ssids iommu/arm-smmu-v3: Add helper for atc invalidation iommu/arm-smmu-v3: Implement set_dev_pasid iommu/arm-smmu-v3-sva: Remove bond refcount iommu/arm-smmu-v3-sva: Clean unused iommu_sva iommu/arm-smmu-v3-sva: Remove arm_smmu_bond iommu/arm-smmu-v3-sva: Add check when enabling sva
.../iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c | 156 +++--- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 447 ++++++++++++------ drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h | 49 +- 3 files changed, 414 insertions(+), 238 deletions(-)
base-commit: b6dad5178ceaf23f369c3711062ce1f2afc33644
Signature · 1 line
2.41.0.162.gfafddb0af9-goog
PATCH v4 1/13 iommu/arm-smmu-v3: Move ctx_desc out of s1_cfg
s1_cfg describes the CD table that is inserted into an SMMU's STEs. It's weird for s1_cfg to also own ctx_desc which describes a CD that is inserted into that table. It is more appropriate for arm_smmu_domain to own ctx_desc.
--- .../iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c | 2 +- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 23 +++++++-------- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h | 28 ++++++++++--------- 3 files changed, 28 insertions(+), 25 deletions(-)
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c +1 -1
@@ -62,7 +62,7 @@ arm_smmu_share_asid(struct mm_struct *mm, u16 asid)return cd; } - smmu_domain = container_of(cd, struct arm_smmu_domain, s1_cfg.cd); + smmu_domain = container_of(cd, struct arm_smmu_domain, cd); smmu = smmu_domain->smmu; ret = xa_alloc(&arm_smmu_asid_xa, &new_asid, cd,
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +12 -11
@@ -1863,7 +1863,7 @@ static void arm_smmu_tlb_inv_context(void *cookie)* careful, 007. */ if (smmu_domain->stage ==ARM_SMMU_DOMAIN_S1 ) { -arm_smmu_tlb_inv_asid (smmu, smmu_domain->s1_cfg.cd.asid); +arm_smmu_tlb_inv_asid (smmu, smmu_domain->cd.asid); }else { cmd.opcode =CMDQ_OP_TLBI_S12_VMALL ; cmd.tlbi.vmid = smmu_domain->s2_cfg.vmid;
@@ -1946,7 +1946,7 @@ static void arm_smmu_tlb_inv_range_domain(unsigned long iova, size_t size,if (smmu_domain->stage ==ARM_SMMU_DOMAIN_S1 ) { cmd.opcode = smmu_domain->smmu->features &ARM_SMMU_FEAT_E2H ?CMDQ_OP_TLBI_EL2_VA :CMDQ_OP_TLBI_NH_VA ; - cmd.tlbi.asid = smmu_domain->s1_cfg.cd.asid; + cmd.tlbi.asid = smmu_domain->cd.asid; }else { cmd.opcode =CMDQ_OP_TLBI_S2_IPA ; cmd.tlbi.vmid = smmu_domain->s2_cfg.vmid;
@@ -2077,7 +2077,7 @@ static void arm_smmu_domain_free(struct iommu_domain *domain)mutex_lock (&arm_smmu_asid_lock);if (cfg->cdcfg.cdtab)arm_smmu_free_cd_tables (smmu_domain); -arm_smmu_free_asid (&cfg->cd); +arm_smmu_free_asid (&smmu_domain->cd);mutex_unlock (&arm_smmu_asid_lock); }else {struct arm_smmu_s2_cfg *cfg = &smmu_domain->s2_cfg;
@@ -2096,13 +2096,14 @@ static int arm_smmu_domain_finalise_s1(struct arm_smmu_domain *smmu_domain,u32 asid;struct arm_smmu_device *smmu = smmu_domain->smmu;struct arm_smmu_s1_cfg *cfg = &smmu_domain->s1_cfg; +struct arm_smmu_ctx_desc *cd = &smmu_domain->cd; typeof(&pgtbl_cfg->arm_lpae_s1_cfg.tcr) tcr = &pgtbl_cfg->arm_lpae_s1_cfg.tcr; -refcount_set (&cfg->cd.refs,1 ); +refcount_set (&cd->refs,1 );/* Prevent SVA from modifying the ASID until it is written to the CD */ mutex_lock (&arm_smmu_asid_lock); - ret =xa_alloc (&arm_smmu_asid_xa, &asid, &cfg->cd, + ret =xa_alloc (&arm_smmu_asid_xa, &asid, cd,XA_LIMIT (1 , (1 << smmu->asid_bits) -1 ),GFP_KERNEL );if (ret) goto out_unlock;
@@ -2115,23 +2116,23 @@ static int arm_smmu_domain_finalise_s1(struct arm_smmu_domain *smmu_domain,if (ret) goto out_free_asid; - cfg->cd.asid = (u16 )asid; - cfg->cd.ttbr = pgtbl_cfg->arm_lpae_s1_cfg.ttbr; - cfg->cd.tcr =FIELD_PREP (CTXDESC_CD_0_TCR_T0SZ , tcr->tsz) | + cd->asid = (u16 )asid; + cd->ttbr = pgtbl_cfg->arm_lpae_s1_cfg.ttbr; + cd->tcr =FIELD_PREP (CTXDESC_CD_0_TCR_T0SZ , tcr->tsz) |FIELD_PREP (CTXDESC_CD_0_TCR_TG0 , tcr->tg) |FIELD_PREP (CTXDESC_CD_0_TCR_IRGN0 , tcr->irgn) |FIELD_PREP (CTXDESC_CD_0_TCR_ORGN0 , tcr->orgn) |FIELD_PREP (CTXDESC_CD_0_TCR_SH0 , tcr->sh) |FIELD_PREP (CTXDESC_CD_0_TCR_IPS , tcr->ips) |CTXDESC_CD_0_TCR_EPD1 |CTXDESC_CD_0_AA64 ; - cfg->cd.mair = pgtbl_cfg->arm_lpae_s1_cfg.mair; + cd->mair = pgtbl_cfg->arm_lpae_s1_cfg.mair;/* * Note that this will end up calling arm_smmu_sync_cd() before * the master has been added to the devices list for this domain. * This isn't an issue because the STE hasn't been installed yet. */ - ret =arm_smmu_write_ctx_desc (smmu_domain,0 , &cfg->cd); + ret =arm_smmu_write_ctx_desc (smmu_domain,0 , cd);if (ret) goto out_free_cd_tables;
@@ -2141,7 +2142,7 @@ static int arm_smmu_domain_finalise_s1(struct arm_smmu_domain *smmu_domain, out_free_cd_tables:arm_smmu_free_cd_tables (smmu_domain); out_free_asid: -arm_smmu_free_asid (&cfg->cd); +arm_smmu_free_asid (cd); out_unlock:mutex_unlock (&arm_smmu_asid_lock);return ret;
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h +15 -13
@@ -593,7 +593,6 @@ struct arm_smmu_ctx_desc_cfg {struct arm_smmu_s1_cfg {struct arm_smmu_ctx_desc_cfg cdcfg; -struct arm_smmu_ctx_desc cd;u8 s1fmt;u8 s1cdmax; };
@@ -707,25 +706,28 @@ enum arm_smmu_domain_stage { };struct arm_smmu_domain { -struct arm_smmu_device *smmu; -struct mutex init_mutex;/* Protects smmu pointer */ +struct arm_smmu_device *smmu; +struct mutex init_mutex;/* Protects smmu pointer */ -struct io_pgtable_ops *pgtbl_ops; -bool stall_enabled; -atomic_t nr_ats_masters; +struct io_pgtable_ops *pgtbl_ops; +bool stall_enabled; +atomic_t nr_ats_masters; -enum arm_smmu_domain_stage stage; +enum arm_smmu_domain_stage stage;union { -struct arm_smmu_s1_cfg s1_cfg; -struct arm_smmu_s2_cfg s2_cfg; +struct { +struct arm_smmu_ctx_desc cd; +struct arm_smmu_s1_cfg s1_cfg; + }; +struct arm_smmu_s2_cfg s2_cfg; }; -struct iommu_domain domain; +struct iommu_domain domain; -struct list_head devices; -spinlock_t devices_lock; +struct list_head devices; +spinlock_t devices_lock; -struct list_head mmu_notifiers; +struct list_head mmu_notifiers; };static inline struct arm_smmu_domain *to_smmu_domain (struct iommu_domain *dom)
Signature · 1 line
2.41.0.162.gfafddb0af9-goog
PATCH v4 2/13 iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
In reply to Michael ShavitExcept for Nested domains, arm_smmu_master will own the STEs that are inserted into the arm_smmu_device's STE table.
--- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 28 +++++++++++++-------- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h | 1 + 2 files changed, 18 insertions(+), 11 deletions(-)
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +17 -11
@@ -1126,15 +1126,16 @@ int arm_smmu_write_ctx_desc(struct arm_smmu_domain *smmu_domain, int ssid,return 0 ; } -static int arm_smmu_alloc_cd_tables (struct arm_smmu_domain *smmu_domain) +static int arm_smmu_init_s1_cfg (struct arm_smmu_master *master, +struct arm_smmu_s1_cfg *cfg) {int ret;size_t l1size;size_t max_contexts; -struct arm_smmu_device *smmu = smmu_domain->smmu; -struct arm_smmu_s1_cfg *cfg = &smmu_domain->s1_cfg; +struct arm_smmu_device *smmu = master->smmu;struct arm_smmu_ctx_desc_cfg *cdcfg = &cfg->cdcfg; + cfg->s1cdmax = master->ssid_bits; max_contexts =1 << cfg->s1cdmax; if (!(smmu->features & ARM_SMMU_FEAT_2_LVL_CDTAB) ||
@@ -1175,12 +1176,11 @@ static int arm_smmu_alloc_cd_tables(struct arm_smmu_domain *smmu_domain)return ret; } -static void arm_smmu_free_cd_tables (struct arm_smmu_domain *smmu_domain) +static void arm_smmu_free_cd_tables (struct arm_smmu_device *smmu, +struct arm_smmu_ctx_desc_cfg *cdcfg) {int i;size_t size, l1size; -struct arm_smmu_device *smmu = smmu_domain->smmu; -struct arm_smmu_ctx_desc_cfg *cdcfg = &smmu_domain->s1_cfg.cdcfg;if (cdcfg->l1_desc) { size =CTXDESC_L2_ENTRIES * (CTXDESC_CD_DWORDS <<3 );
@@ -2076,7 +2076,7 @@ static void arm_smmu_domain_free(struct iommu_domain *domain)/* Prevent SVA from touching the CD while we're freeing it */ mutex_lock (&arm_smmu_asid_lock);if (cfg->cdcfg.cdtab) -arm_smmu_free_cd_tables (smmu_domain); +arm_smmu_free_cd_tables (smmu, &cfg->cdcfg);arm_smmu_free_asid (&smmu_domain->cd);mutex_unlock (&arm_smmu_asid_lock); }else {
@@ -2108,11 +2108,9 @@ static int arm_smmu_domain_finalise_s1(struct arm_smmu_domain *smmu_domain,if (ret) goto out_unlock; - cfg->s1cdmax = master->ssid_bits; - smmu_domain->stall_enabled = master->stall_enabled; - ret =arm_smmu_alloc_cd_tables (smmu_domain); + ret =arm_smmu_init_s1_cfg (master, cfg);if (ret) goto out_free_asid;
@@ -2140,7 +2138,7 @@ static int arm_smmu_domain_finalise_s1(struct arm_smmu_domain *smmu_domain,return 0 ; out_free_cd_tables: -arm_smmu_free_cd_tables (smmu_domain); +arm_smmu_free_cd_tables (smmu, &cfg->cdcfg); out_free_asid:arm_smmu_free_asid (cd); out_unlock:
@@ -2704,6 +2702,13 @@ static struct iommu_device *arm_smmu_probe_device(struct device *dev) smmu->features & ARM_SMMU_FEAT_STALL_FORCE) master->stall_enabled = true; + ret =arm_smmu_init_s1_cfg (master, &master->owned_s1_cfg); +if (ret) { +arm_smmu_disable_pasid (master); +arm_smmu_remove_master (master); + goto err_free_master; + } +return &smmu->iommu; err_free_master:
@@ -2719,6 +2724,7 @@ static void arm_smmu_release_device(struct device *dev)if (WARN_ON (arm_smmu_master_sva_enabled (master)))iopf_queue_remove_device (master->smmu->evtq.iopf, dev);arm_smmu_detach_dev (master); +arm_smmu_free_cd_tables (master->smmu, &master->owned_s1_cfg.cdcfg);arm_smmu_disable_pasid (master);arm_smmu_remove_master (master);kfree (master);
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h +1 -0
@@ -688,6 +688,7 @@ struct arm_smmu_master {struct arm_smmu_domain *domain;struct list_head domain_head;struct arm_smmu_stream *streams; +struct arm_smmu_s1_cfg owned_s1_cfg;unsigned int num_streams;bool ats_enabled;bool stall_enabled;
Signature · 1 line
2.41.0.162.gfafddb0af9-goog
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Hi Michael,
Quoted 2 lines · Michael Shavit
Except for Nested domains,
arm_smmu_masterwill own the STEs that are inserted into the arm_smmu_device's STE table.
I think that the master still owns an STE when attached to a nested domain. Though an IOMMU_DOMAIN_NESTED iommu_domain is an opaque object to the STE in the guest, the host still has a real STE for the nested configuration somewhere -- and it's likely still to be owned by the master that's attached to the opaque NESTED iommu_domain in the host kernel.
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c int arm_smmu_write_ctx_desc(struct arm_smmu_domain *smmu_domain, int ssid, · line 1130
Quoted 3 lines
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +2
-1-static int arm_smmu_alloc_cd_tables (struct arm_smmu_domain *smmu_domain) +static int arm_smmu_init_s1_cfg (struct arm_smmu_master *master, +struct arm_smmu_s1_cfg *cfg)
We here pass in an s1_cfg ptr because we expect someone else rather than the master could own the s1_cfg?
But the final codeline by the end of this series seems that only master owns an s1_cfg. So perhaps we could re-organize the patches to clean this away, as the cfg always comes from a master?
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h struct arm_smmu_master { · line 691
Quoted 9 lines
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h +1
-0@@ -688,6 +688,7 @@ struct arm_smmu_master {struct arm_smmu_domain *domain;struct list_head domain_head;struct arm_smmu_stream *streams; +struct arm_smmu_s1_cfg owned_s1_cfg;
I am a bit confused by this naming. If only master would own an s1_cfg, perhaps we can just make it "s1_cfg" and drop the s1_cfg pointer in the next patch.
Thanks
Nicolin
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Quoted 11 lines · Nicolin Chen
Quoted 3 lines
Except for Nested domains,
arm_smmu_masterwill own the STEs that are inserted into the arm_smmu_device's STE table.I think that the master still owns an STE when attached to a nested domain. Though an
IOMMU_DOMAIN_NESTEDiommu_domainis an opaque object to the STE in the guest, the host still has a real STE for the nested configuration somewhere -- and it's likely still to be owned by the master that's attached to the opaque NESTEDiommu_domainin the host kernel.
Quoted 3 lines
I am a bit confused by this naming. If only master would own an
s1_cfg, perhaps we can just make it "s1_cfg" and drop thes1_cfgpointer in the next patch.
Could be that the naming is causing some confusion. This owned_s1_cfg is very different from the s1_cfg set-up by Nested domains in your patch series. It's better to think of it as the default s1_cfg used for DMA/SVA/UNMANAGED domains. Because stage 1 domains represent a single page table, it doesn't make sense for them to own an entire CD table. In contrast, nested domains map an entire CD table and it therefore makes sense for them to own the s1_cfg representing that table.
Would renaming this as default_s1_cfg make more sense?
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Quoted 16 lines · Michael Shavit
Quoted 11 lines · Nicolin Chen
Quoted 3 lines
Except for Nested domains,
arm_smmu_masterwill own the STEs that are inserted into the arm_smmu_device's STE table.I think that the master still owns an STE when attached to a nested domain. Though an
IOMMU_DOMAIN_NESTEDiommu_domainis an opaque object to the STE in the guest, the host still has a real STE for the nested configuration somewhere -- and it's likely still to be owned by the master that's attached to the opaque NESTEDiommu_domainin the host kernel.Quoted 3 lines
I am a bit confused by this naming. If only master would own an
s1_cfg, perhaps we can just make it "s1_cfg" and drop thes1_cfgpointer in the next patch.
Quoted 9 lines
Could be that the naming is causing some confusion. This
owned_s1_cfgis very different from thes1_cfgset-up by Nested domains in your patch series. It's better to think of it as the defaults1_cfgused for DMA/SVA/UNMANAGED domains. Because stage 1 domains represent a single page table, it doesn't make sense for them to own an entire CD table. In contrast, nested domains map an entire CD table and it therefore makes sense for them to own thes1_cfgrepresenting that table.
Would renaming this asdefault_s1_cfgmake more sense?
It would make alot more sense if the STE value used by an unmanaged S1 domain was located in/near the unmanaged domain or called 'unmanaged S1 STE' or something if it really has to be in the master. Why does this even need to be stored, can't we compute it?
Notice that we have unmanaged domains that use a CD and SVA domains typically would be on a CD, so it is a bit weird already.
I'd think the basic mental model should be to extract the STE from the thing you intend to install. Either the default CD table, or from the iommu_domain. ie some 'get STE from iommu_domain' function?
There shouldn't be a concept of a "default" or "owned" STE value..
Jason
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Quoted 4 lines · Jason Gunthorpe
It would make alot more sense if the STE value used by an unmanaged S1 domain was located in/near the unmanaged domain or called 'unmanaged S1 STE' or something if it really has to be in the master. Why does this even need to be stored, can't we compute it?
struct s1_cfg* and struct s2_cfg* are precisely what is used to compute an STE. For example, when s1_cfg is set, arm_smmu_write_strtab will write the s1_cfg's CD table dma_pointer into the STE's STRTAB_STE_0_CFG field. When neither are set, the STE fields are written to enable bypass (or abort depending on the config).
Quoted 3 lines
I'd think the basic mental model should be to extract the STE from the thing you intend to install. Either the default CD table, or from the
iommu_domain. ie some 'get STE fromiommu_domain' function?
I don't follow this. When we attach a domain with pasid (whether through SVA or the set_dev_pasid API) , we don't want to install an entirely new CD table. We want to write something (page-table pointer) to a common CD table. Where should the s1_cfg which owns that common table live? I thought we concluded that it should be owned by the arm_smmu_master rather than any domain (to avoid dependencies between domains a-la aux-domain). With this change, even attach_dev with a DMA or UNMANAGED domain is now just preparing a single entry into this common CD table.
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Quoted 3 lines · Michael Shavit
domains a-la aux-domain). With this change, even
attach_devwith a DMA or UNMANAGED domain is now just preparing a single entry into this common CD table.
I have to correct this part, arm-smmu-v3.c's attach_dev() implementation does both: it writes to the pasid=0 entry of the owned_s1_cfg's CD table, and then installs that CD table to the STE.
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
In reply to Michael ShavitQuoted 11 lines · Michael Shavit
Quoted 4 lines · Jason Gunthorpe
It would make alot more sense if the STE value used by an unmanaged S1 domain was located in/near the unmanaged domain or called 'unmanaged S1 STE' or something if it really has to be in the master. Why does this even need to be stored, can't we compute it?
struct s1_cfg* and struct s2_cfg* are precisely what is used to compute an STE. For example, when
s1_cfgis set,arm_smmu_write_strtabwill write the s1_cfg's CD tabledma_pointerinto the STE'sSTRTAB_STE_0_CFGfield. When neither are set, the STE fields are written to enable bypass (or abort depending on the config).
I guess I never really understood why these were precomputed and stored at all. Even more confusing is why we need to keep pointers to them anywhere? Compute the STE and CDE directly from the source data when you need it?
eg If I want to install an IDENITY domain into a STE then I compute the STE for identity and go ahead and do it.
Quoted 7 lines
Quoted 3 lines
I'd think the basic mental model should be to extract the STE from the thing you intend to install. Either the default CD table, or from the
iommu_domain. ie some 'get STE fromiommu_domain' function?I don't follow this. When we attach a domain with pasid (whether through SVA or the
set_dev_pasidAPI) , we don't want to install an entirely new CD table.
The master object owns an optional CD table. If it is exsists it is used by every domain that is attached to that master.
In the code flow there are two entry points to attach a domain, attach to a PASID or attach to a RID.
For attach to PASID the code should always force the master to have a CD table and then attach the domain to the CD table.
For attach to RID the code should do a bunch of checks and decide if it should force the master to have a CD table and attach the domain to that, or directly attach the domain to the STE.
When the master gains a CD table then the CD table object becomes attached to the STE. In all cases we should be able to point to the object the STE points at and don't need a cfg or pointer to cfg since the object itself can provide the cfg.
In all cases when you go to compute a STE you figure out what object is attached to it (CD or domain), compute the correct STE for that object, the set it. Same for he CDE, query the correct CDE from the iommu_domain when you attach it to the table.
There should be no such thing as a "default" STE, and I question if it makes sense to even precompute the s1/s2_cfg values during finalize at all..
Quoted 2 lines
We want to write something (page-table pointer) to a common CD table. Where should the
s1_cfgwhich owns that common table live?
I would suggest a 'cd table struct' that as all the stuff related to the CD table, including an API to cacluate the STE this CD table requires. If not in actual code with a real struct, then in a logical sense in that a chunk of the master struct is the "CD table".
Quoted 3 lines
I thought we concluded that it should be owned by the
arm_smmu_masterrather than any domain (to avoid dependencies between domains a-la aux-domain).
Yes, I'm not saying anything against that, just how and where the STE and CDE values flow around.
Jason
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Quoted 13 lines · Jason Gunthorpe
Quoted 11 lines · Michael Shavit
Quoted 4 lines · Jason Gunthorpe
It would make alot more sense if the STE value used by an unmanaged S1 domain was located in/near the unmanaged domain or called 'unmanaged S1 STE' or something if it really has to be in the master. Why does this even need to be stored, can't we compute it?
struct s1_cfg* and struct s2_cfg* are precisely what is used to compute an STE. For example, when
s1_cfgis set,arm_smmu_write_strtabwill write the s1_cfg's CD tabledma_pointerinto the STE'sSTRTAB_STE_0_CFGfield. When neither are set, the STE fields are written to enable bypass (or abort depending on the config).
Quoted 7 lines
I guess I never really understood why these were precomputed and stored at all. Even more confusing is why we need to keep pointers to them anywhere? Compute the STE and CDE directly from the source data when you need it?
eg If I want to install an IDENITY domain into a STE then I compute the STE for identity and go ahead and do it.
I think it'd be clear if the master stores STE value directly to get rid of s1_cfg/s2_cfg in the master struct. We fill in those domain-related STE fields when the domain attaches to the master and then call arm_smmu_write_strtab().
For struct arm_smmu_domain, it still has a union of a CD (for an S1 domain) or an s2_cfg (for an S2 domain). Or we could select a better naming for s2_cfg too, since s1_cfg is gone.
Does this match with your expectation?
Quoted 14 lines
Quoted 7 lines
Quoted 3 lines
I'd think the basic mental model should be to extract the STE from the thing you intend to install. Either the default CD table, or from the
iommu_domain. ie some 'get STE fromiommu_domain' function?I don't follow this. When we attach a domain with pasid (whether through SVA or the
set_dev_pasidAPI) , we don't want to install an entirely new CD table.The master object owns an optional CD table. If it is exsists it is used by every domain that is attached to that master.
In the code flow there are two entry points to attach a domain, attach to a PASID or attach to a RID.
Quoted 6 lines
For attach to PASID the code should always force the master to have a CD table and then attach the domain to the CD table.
For attach to RID the code should do a bunch of checks and decide if it should force the master to have a CD table and attach the domain to that, or directly attach the domain to the STE.
I think a master own a CD table in both cases, though the table could have a single entry or multiple entries, depending on its ssid_bits/cdmax value. And since a master own the CD table, we could preallocate the CD table in arm_smmu_probe_device(), like Michael does in this series. And at the same time, the s1ctxptr field of the STE could be setup too. Then we insert the CD from a domain into the CD table during the domain attaching.
Yet two cases that would waste the preallocated CD table:
1) If a master with ssid_bits=0 gets attached to an IDENITY S1 domain, it sets CONFIG=BYPASS in the STE, which wastes the single-entry CD table, unlike currently the driver bypassing the allocation of a CD table at all.
2) When enabling nested translation, we should replace all the S1 fields in the STE with guest/user values. So, the kernel- level CD table (either single-entry or multi-entry) in the master struct will not be used. Although this seems to be unchanged since currently the driver wastes this too in the default domain?
With that, I still feel it clear and flexible. And I can simply add my use case of attaching an IDENITY domain to the default substream when having a multi-entry CD table.
Thanks
Nicolin
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Quoted 19 lines · Nicolin Chen
Quoted 18 lines · Jason Gunthorpe
Quoted 11 lines · Michael Shavit
Quoted 4 lines · Jason Gunthorpe
It would make alot more sense if the STE value used by an unmanaged S1 domain was located in/near the unmanaged domain or called 'unmanaged S1 STE' or something if it really has to be in the master. Why does this even need to be stored, can't we compute it?
struct s1_cfg* and struct s2_cfg* are precisely what is used to compute an STE. For example, when
s1_cfgis set,arm_smmu_write_strtabwill write the s1_cfg's CD tabledma_pointerinto the STE'sSTRTAB_STE_0_CFGfield. When neither are set, the STE fields are written to enable bypass (or abort depending on the config).I guess I never really understood why these were precomputed and stored at all. Even more confusing is why we need to keep pointers to them anywhere? Compute the STE and CDE directly from the source data when you need it?
Quoted 7 lines
Quoted 2 lines
eg If I want to install an IDENITY domain into a STE then I compute the STE for identity and go ahead and do it.
I think it'd be clear if the master stores STE value directly to get rid of s1_cfg/s2_cfg in the master struct. We fill in those domain-related STE fields when the domain attaches to the master and then call
arm_smmu_write_strtab().
Right, though master effectively records the STE when it writes it to the steering table. If we need another copy I don't know.
Quoted 3 lines
For struct
arm_smmu_domain, it still has a union of a CD (for an S1 domain) or ans2_cfg(for an S2 domain). Or we could select a better naming fors2_cfgtoo, sinces1_cfgis gone.
By spec it should be called CD and STE value, but frankly I like CDTE and STE a little better (context descriptor table entry) as it evokes a sense of similarity. I don't care for the words 's1_cfg' and 's2_cfg' to represent the CD and STE, that is pretty confusing now that we have more uses for the CD and STE's than simple s1/s2 IO page tables.
Quoted 2 lines
I think a master own a CD table in both cases, though the table could have a single entry or multiple entries
Yes.
Quoted 6 lines
depending on its ssid_bits/cdmax value. And since a master own the CD table, we could preallocate the CD table in
arm_smmu_probe_device(), like Michael does in this series. And at the same time, the s1ctxptr field of the STE could be setup too. Then we insert the CD from a domain into the CD table during the domain attaching.
We insert the CDTE from the domain into the CD table, and generate a STE for the CD table and insert it ino the Steering table.
Quoted 11 lines
Yet two cases that would waste the preallocated CD table:
1) If a master with ssid_bits=0 gets attached to an IDENITY S1 domain, it sets CONFIG=BYPASS in the STE, which wastes the single-entry CD table, unlike currently the driver bypassing the allocation of a CD table at all.
2) When enabling nested translation, we should replace all the S1 fields in the STE with guest/user values. So, the kernel- level CD table (either single-entry or multi-entry) in the master struct will not be used. Although this seems to be unchanged since currently the driver wastes this too in the default domain?
As we discussed in another thread dynamic resizing of the CD table makes some sense. Eg keep it at one entry until PASID mode is enabled then resize it to the max PASID bits.
But I view that as an improvement beyond here. It seems fine for now to pre allocate the full CD table and waste the memory. PASID capable devices are rare.
Quoted 3 lines
With that, I still feel it clear and flexible. And I can simply add my use case of attaching an IDENITY domain to the default substream when having a multi-entry CD table.
Yes, with the above note about dynamically resizing the CD table, I think we generally have the idea that programming the CD, eg by installing an entry, can cause the CD tables's STE to (atomically) change.
Eg to toggle the S1DSS bit if the PASID 0 is IDENTITY, or to resize the table if we exceed the PASID space.
Longer term we'd also need a way to setup IDENTITY domains for !0 substreams as well too (eg for SIOV). It is too bad the CDTE doesn't have a bit for bypass. I suppose it will need dummy 1:1 IO page tables or something.
Jason
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Quoted 10 lines · Jason Gunthorpe
Quoted 3 lines
For struct
arm_smmu_domain, it still has a union of a CD (for an S1 domain) or ans2_cfg(for an S2 domain). Or we could select a better naming fors2_cfgtoo, sinces1_cfgis gone.By spec it should be called CD and STE value, but frankly I like CDTE and STE a little better (context descriptor table entry) as it evokes a sense of similarity. I don't care for the words '
s1_cfg' and 's2_cfg' to represent the CD and STE, that is pretty confusing now that we have more uses for the CD and STE's than simple s1/s2 IO page tables.
We have STE and PTE, so CDTE sounds legit to me, though probably it'd be safer by just following the spec? We can always add kdoc and comments to make things clear.
@Michael,
Would it be possible for you to update a v5, following Jason's suggestion overall? And I think we can have a smaller refactor series first without set_dev_pasid, to have a common codeline sooner for us to add new features, such as set_dev_pasid, the use case of IDENTITY default substream, and the nesting series. I will help testing with some pasid/non-pasid use cases too.
Quoted 12 lines
Quoted 11 lines
Yet two cases that would waste the preallocated CD table:
1) If a master with ssid_bits=0 gets attached to an IDENITY S1 domain, it sets CONFIG=BYPASS in the STE, which wastes the single-entry CD table, unlike currently the driver bypassing the allocation of a CD table at all.
2) When enabling nested translation, we should replace all the S1 fields in the STE with guest/user values. So, the kernel- level CD table (either single-entry or multi-entry) in the master struct will not be used. Although this seems to be unchanged since currently the driver wastes this too in the default domain?
Quoted 3 lines
As we discussed in another thread dynamic resizing of the CD table makes some sense. Eg keep it at one entry until PASID mode is enabled then resize it to the max PASID bits.
I see.
Quoted 3 lines
But I view that as an improvement beyond here. It seems fine for now to pre allocate the full CD table and waste the memory. PASID capable devices are rare.
Yea, that'd be easier for us to move forward other features :)
Quoted 11 lines
Quoted 3 lines
With that, I still feel it clear and flexible. And I can simply add my use case of attaching an IDENITY domain to the default substream when having a multi-entry CD table.
Yes, with the above note about dynamically resizing the CD table, I think we generally have the idea that programming the CD, eg by installing an entry, can cause the CD tables's STE to (atomically) change.
Eg to toggle the S1DSS bit if the PASID 0 is IDENTITY, or to resize the table if we exceed the PASID space.
Yea, we have enable_pasid() so probably can resize over there.
Quoted 4 lines
Longer term we'd also need a way to setup IDENTITY domains for !0 substreams as well too (eg for SIOV). It is too bad the CDTE doesn't have a bit for bypass. I suppose it will need dummy 1:1 IO page tables or something.
Oh. I haven't thought about that far. Noted it down.
Thanks
Nicolin
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Quoted 7 lines · Nicolin Chen
@Michael,
Would it be possible for you to update a v5, following Jason's suggestion overall? And I think we can have a smaller refactor series first withoutset_dev_pasid, to have a common codeline sooner for us to add new features, such asset_dev_pasid, the use case of IDENTITY default substream, and the nesting series. I will help testing with some pasid/non-pasid use cases too.
Want to make sure I fully understand these last few messages first. At a high level, we want:
1. arm_smmu_master is allowed to own a CD table, but not an STE-precursor (s1_cfg/s2_cfg). The s1_cfg is practically already that, and we can probably get there with minimal changes.
2. arm_smmu_master shouldn't point to the currently active CD table (which may or may not be the one it owns) or STE-precursor as a shortcut. All code should figure it out by looking at the master's currently attached domain (functionality could be provided by helper).
3. arm_smmu_domain shouldn't pre-generate any STE-precursors. The STE/CD for a domain should either be computed right when it is written, or computed ahead of time and stored as a copy in the smmu-domain.
Is that right?
Will Deacon
maintainerRe: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Quoted 9 lines · Michael Shavit
Quoted 7 lines · Nicolin Chen
@Michael,
Would it be possible for you to update a v5, following Jason's suggestion overall? And I think we can have a smaller refactor series first withoutset_dev_pasid, to have a common codeline sooner for us to add new features, such asset_dev_pasid, the use case of IDENTITY default substream, and the nesting series. I will help testing with some pasid/non-pasid use cases too.
Quoted 14 lines
Want to make sure I fully understand these last few messages first. At a high level, we want:
1.arm_smmu_masteris allowed to own a CD table, but not an STE-precursor (s1_cfg/s2_cfg). Thes1_cfgis practically already that, and we can probably get there with minimal changes.
2.arm_smmu_mastershouldn't point to the currently active CD table (which may or may not be the one it owns) or STE-precursor as a shortcut. All code should figure it out by looking at the master's currently attached domain (functionality could be provided by helper).
3.arm_smmu_domainshouldn't pre-generate any STE-precursors. The STE/CD for a domain should either be computed right when it is written, or computed ahead of time and stored as a copy in the smmu-domain.
Is that right?
To be honest, I'm having a hard time following the thread. I'd like to avoid renaming things "for fun", so let's try to focus on the actual restructuring initially. I'd also like to see what this "computing" of the SMMU data structures actually looks like -- the information stored in the 'arm_smmu_master' structure isn't redundant, so what's the problem?
If you all grok this, then I'm happy to wait for the patches, but I don't want you to go away and spend lots of effort hacking something which doesn't end up being what folks are asking for.
Will
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
In reply to Michael ShavitQuoted 9 lines · Michael Shavit
Quoted 7 lines · Nicolin Chen
@Michael,
Would it be possible for you to update a v5, following Jason's suggestion overall? And I think we can have a smaller refactor series first withoutset_dev_pasid, to have a common codeline sooner for us to add new features, such asset_dev_pasid, the use case of IDENTITY default substream, and the nesting series. I will help testing with some pasid/non-pasid use cases too.
Quoted 5 lines
Want to make sure I fully understand these last few messages first. At a high level, we want:
1.arm_smmu_masteris allowed to own a CD table, but not an STE-precursor (s1_cfg/s2_cfg). Thes1_cfgis practically already that, and we can probably get there with minimal changes.
Yah
Quoted 5 lines
2.
arm_smmu_mastershouldn't point to the currently active CD table (which may or may not be the one it owns) or STE-precursor as a shortcut. All code should figure it out by looking at the master's currently attached domain (functionality could be provided by helper).
I think that is close.
Most likely the master should have a pointer to an STE owning iommu_domain. If that is !NULL then the master's CD table is not used and the STE is computed from that iommu_domain. Essentially it says that iommu_domain owns the entire RID.
Otherwise the STE comes from the master's CD table, and this means the master is in SSID mode.
Quoted 4 lines
3.
arm_smmu_domainshouldn't pre-generate any STE-precursors. The STE/CD for a domain should either be computed right when it is written, or computed ahead of time and stored as a copy in the smmu-domain.
I think this is cleaner since alot of the STE/CD calculation is either constant or copying data from other places in the domain struct, but maybe you'll find this is wrong. At least it is not super important at this point.
Jason
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
In reply to Jason GunthorpeQuoted 6 lines · Jason Gunthorpe
The master object owns an optional CD table. If it is exsists it is used by every domain that is attached to that master.
In the code flow there are two entry points to attach a domain, attach to a PASID or attach to a RID.
Quoted 6 lines
For attach to PASID the code should always force the master to have a CD table and then attach the domain to the CD table.
For attach to RID the code should do a bunch of checks and decide if it should force the master to have a CD table and attach the domain to that, or directly attach the domain to the STE.
Yes. This is the current flow (except that we fail instead of forcing
when a CD table isn't already attached in the PASID flow). owned_s1_cfg is simply a pre-allocated version of your optional CD table.
Quoted 4 lines
When the master gains a CD table then the CD table object becomes attached to the STE. In all cases we should be able to point to the object the STE points at and don't need a cfg or pointer to cfg since the object itself can provide the cfg.
Ok, practically speaking, are we just talking about reverting patch 3 and keeping a handle to the primary domain in arm_smmu_master? Instead of directly accessing the currently active CD table using arm_smmu_master->s1_cfg, you'd like set_dev_pasid() to look for it by investigating what kind of domain is attached, and reaching to the pre-alllocated/optional CD table otherwise if necessary?
Quoted 7 lines
Quoted 2 lines
We want to write something (page-table pointer) to a common CD table. Where should the
s1_cfgwhich owns that common table live?I would suggest a 'cd table struct' that as all the stuff related to the CD table, including an API to cacluate the STE this CD table requires. If not in actual code with a real struct, then in a logical sense in that a chunk of the master struct is the "CD table".
Sure, that's almost exactly what s1_cfg is today (with these patches)....
* s1_cfg.arm_smmu_ctx_desc_cfg describes the CD table
* s1_cfg.s1fmt and s1_cfg.s1cdmax describes attributes of that CD
table. These could technically be deduced-back from arm_smmu_ctx_desc_cfg's l1_desc and num_l1_ents
* s1_cfg.stall_enabled was introduced by patch 4 and in retrospect
does not belong there at all. It should have gone in arm_smmu_ctx_desc instead (ignoring the whole should arm_smmu_ctx_desc even be precomputed in the first place topic for a second).
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Quoted 10 lines · Michael Shavit
Quoted 9 lines · Jason Gunthorpe
The master object owns an optional CD table. If it is exsists it is used by every domain that is attached to that master.
In the code flow there are two entry points to attach a domain, attach to a PASID or attach to a RID.
For attach to PASID the code should always force the master to have a CD table and then attach the domain to the CD table.
Quoted 8 lines
Quoted 3 lines
For attach to RID the code should do a bunch of checks and decide if it should force the master to have a CD table and attach the domain to that, or directly attach the domain to the STE.
Yes. This is the current flow (except that we fail instead of forcing
when a CD table isn't already attached in the PASID flow).owned_s1_cfgis simply a pre-allocated version of your optional CD table.
Really? That seems like a terrible name for the CD table.
Quoted 7 lines
Quoted 4 lines
When the master gains a CD table then the CD table object becomes attached to the STE. In all cases we should be able to point to the object the STE points at and don't need a cfg or pointer to cfg since the object itself can provide the cfg.
Ok, practically speaking, are we just talking about reverting patch 3 and keeping a handle to the primary domain in
arm_smmu_master?
I think the master should have a pointer to the iommu_domain that owns the STE or if NULL the master should assign its internal CD table to the STE.
The iommu_domain structs should not have any references to a CD table.
Quoted 10 lines
Quoted 4 lines
I would suggest a 'cd table struct' that as all the stuff related to the CD table, including an API to cacluate the STE this CD table requires. If not in actual code with a real struct, then in a logical sense in that a chunk of the master struct is the "CD table".
Sure, that's almost exactly what
s1_cfgis today (with these patches)....
* s1_cfg.arm_smmu_ctx_desc_cfg describes the CD table
* s1_cfg.s1fmt and s1_cfg.s1cdmax describes attributes of that CD
table. These could technically be deduced-back from arm_smmu_ctx_desc_cfg'sl1_descandnum_l1_ents
Yes, this makes sense, there is some redundancy here for sure.
patch 1 makes sense, arm_smmu_ctx_desc is effectively the Context Descriptor Entry, and it belongs in the domain
patch 2 should delete arm_smmu_s1_cfg and just put arm_smmu_ctx_desc_cfg directly in the master. arm_smmu_ctx_desc_cfg is a weird name for the contex descriptor table, but it is much less weird than s1_cfg. As you say s1fmt/s1cdmax are redundant.
patch 3 I don't understand, we should not add something called s1_cfg/s2_cfg to the master. The master should have 'arm_smmu_ctx_desc_cfg cd_table' and 'arm_smmu_domain ste_domain'
patch 4 should have everything working with the cd table accept a 'struct arm_smmu_ctx_desc_cfg *', eg arm_smmu_get_cd_ptr (get a pointer to a CD entry in the CD table).
patch 5 makes sense, but something seems odd about the order as we somehow half moved it in patch 2?
My suggestion for patch structure is to start by cleaning up the CD table object. Make arm_smmu_ctx_desc_cfg the CD table, remove the redudencies, remove arm_s1_cfg, clean the CD table APIs to only use 'struct arm_smmu_ctx_desc_cfg *', add the 'ste_domain' to the master, and then as the last step just move the arm_smmu_ctx_desc_cfg from the iommu_domain to the master.
And that is a nice little series on its own - you end up with a shared CD table in the master, and no CD table in any domains.
Jason
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Quoted 4 lines · Jason Gunthorpe
patch 2 should delete
arm_smmu_s1_cfgand just putarm_smmu_ctx_desc_cfgdirectly in the master.arm_smmu_ctx_desc_cfgis a weird name for the contex descriptor table, but it is much less weird thans1_cfg. As you say s1fmt/s1cdmax are redundant.
s1fmt is fairly trivial to replace but s1cdmax requires inversing previous computations. I don't really buy that getting rid of it simplifies anything, even if it's technically redundant.
Quoted 3 lines
patch 3 I don't understand, we should not add something called s1_cfg/s2_cfg to the master. The master should have '
arm_smmu_ctx_desc_cfgcd_table' and 'arm_smmu_domainste_domain'
This was simply meant to be a more convenient way of finding the currently active cdtable from the arm_smmu_write_ctx_desc/write_strtab_ent functions without having to inspect the currently attached domain. But sure, that's easy enough to revert.
Quoted 2 lines
patch 5 makes sense, but something seems odd about the order as we somehow half moved it in patch 2?
Ack; patch 2 can be reordered to come after patch 4, or even squashed with 5 if you prefer.
Quoted 9 lines
My suggestion for patch structure is to start by cleaning up the CD table object. Make
arm_smmu_ctx_desc_cfgthe CD table, remove the redudencies, removearm_s1_cfg, clean the CD table APIs to only use 'structarm_smmu_ctx_desc_cfg*', add the 'ste_domain' to the master, and then as the last step just move thearm_smmu_ctx_desc_cfgfrom theiommu_domainto the master.And that is a nice little series on its own - you end up with a shared CD table in the master, and no CD table in any domains.
I don't entirely buy that refactoring s1_cfg is worth the extra effort, nor that it should be tied to this patch series. This series already makes s1_cfg behave as the CD table; whether we want to entirely get rid of pre-computed data useful for writing an STE sounds like a separate cleanup.
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Quoted 9 lines · Michael Shavit
Quoted 4 lines · Jason Gunthorpe
patch 2 should delete
arm_smmu_s1_cfgand just putarm_smmu_ctx_desc_cfgdirectly in the master.arm_smmu_ctx_desc_cfgis a weird name for the contex descriptor table, but it is much less weird thans1_cfg. As you say s1fmt/s1cdmax are redundant.s1fmt is fairly trivial to replace but s1cdmax requires inversing previous computations. I don't really buy that getting rid of it simplifies anything, even if it's technically redundant.
Then store s1cdmax in the arm_smmu_ctx_desc_cfg and still get rid of arm_smmu_s1_cfg
The point is to get rid of "s1_cfg" entirely as language in the driver *because it makes zero sense now*. Prior to your restructuring it was sort of the STE to use for the S1 page table which had an embedded CD. After all this change it isn't realated to a S1 page table anymore at all.
Quoted 6 lines
Quoted 3 lines
patch 3 I don't understand, we should not add something called s1_cfg/s2_cfg to the master. The master should have '
arm_smmu_ctx_desc_cfgcd_table' and 'arm_smmu_domainste_domain'This was simply meant to be a more convenient way of finding the currently active cdtable from the
There is only one (meaningful) choice though, the cdtable is either the master's default CD table or it isn't. You detect that by checking for NULL ste_domain
Quoted 7 lines
Quoted 7 lines
My suggestion for patch structure is to start by cleaning up the CD table object. Make
arm_smmu_ctx_desc_cfgthe CD table, remove the redudencies, removearm_s1_cfg, clean the CD table APIs to only use 'structarm_smmu_ctx_desc_cfg*', add the 'ste_domain' to the master, and then as the last step just move thearm_smmu_ctx_desc_cfgfrom theiommu_domainto the master.
Quoted 8 lines
Quoted 2 lines
And that is a nice little series on its own - you end up with a shared CD table in the master, and no CD table in any domains.
I don't entirely buy that refactoring
s1_cfgis worth the extra effort, nor that it should be tied to this patch series. This series already makess1_cfgbehave as the CD table; whether we want to entirely get rid of pre-computed data useful for writing an STE sounds like a separate cleanup.
s1_cfg is a terrible name to describe the cd table. Please do the tiny bit extra to get rid of it for clarity.
Jason
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Haven't had time to address this yet; but Ack on last message. Will address in a v5 series.
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Sorry for the delay; I'm trying to refactor these patches now.
Quoted 3 lines
I think the master should have a pointer to the
iommu_domainthat owns the STE or if NULL the master should assign its internal CD table to the STE.
Just to clarify, does the nested domain patch series require writing CDs into the user-space's CD table using arm_smmu_write_ctx_desc()? Or is there any other requirement for writing a CD into a domain-owned CD table from arm_smmu_write_ctx_desc?
Writing a CD entry to the CD table involves some dependencies on the master(s) that the table is attached to:
1. The CD entries STALL bit value in arm_smmu_write_ctx_desc depends on the master (e.g. if STALL_FORCE is set on the smmu device). This could potentially be encoded in arm_smmu_ctx_desc_cfg, at which point that CD table is only attachable to masters with the same stall_enabled value.
2. arm_smmu_write_ctx_desc must sync the CD for the attached master(s) in the middle of writing CD entry.
This is all easier to handle in arm_smmu_write_ctx_desc if the table is always owned by the master.
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Quoted 7 lines · Michael Shavit
Sorry for the delay; I'm trying to refactor these patches now.
Quoted 3 lines
I think the master should have a pointer to the
iommu_domainthat owns the STE or if NULL the master should assign its internal CD table to the STE.Just to clarify, does the nested domain patch series require writing CDs into the user-space's CD table using
arm_smmu_write_ctx_desc()?
No, CD entries in nested CD tables are written by userspace only.
Quoted 2 lines
Or is there any other requirement for writing a CD into a domain-owned CD table from
arm_smmu_write_ctx_desc?
Not that I know of.
The only time the kernel writes a CD entry is to the shared CD table stored in the master.
Quoted 5 lines
1. The CD entries STALL bit value in
arm_smmu_write_ctx_descdepends on the master (e.g. ifSTALL_FORCEis set on the smmu device). This could potentially be encoded inarm_smmu_ctx_desc_cfg, at which point that CD table is only attachable to masters with the samestall_enabledvalue.
For cleanness I would orgnize it like this.
Quoted 5 lines
2.
arm_smmu_write_ctx_descmust sync the CD for the attached master(s) in the middle of writing CD entry.This is all easier to handle in
arm_smmu_write_ctx_descif the table is always owned by the master.
I think it is fine if you start with a shared CD table being 1:1 with a single master.
Making the CD table shared between masters (eg for multi-device-group) is a micro-optimization, and I'm not sure we have workloads where it is worthwhile. We already block PASID support for multi-device-group.
Though resizable CD table is probably a better place to put efforts.
Jason
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Awesome that helps a lot. These are the patches I have in mind:
1. Move ctx_desc out of s1_cfg
2. Replace domain->s1_cfg with struct arm_smmu_ctx_desc_cfg cd_table
3. Move stall_enabled from domain to arm_smmu_ctx_desc_cfg
4. Accept an arm_smmu_master instead of an arm_smmu_domain as the parameter for arm_smmu_write_ctx_desc
--- arm_smmu_write_ctx_desc can simply get the cd_table it writes to from master->domain->cd_table; this get's replaced by master->cd_table in subsequent commits.
--- SVA is weird if we cut changes here: it iterates over all masters a domain is attached to in order to call arm_smmu_write_ctx_desc(), which ends up writing to the shared cd_table since in master->domain. This becomes more sensible once masters stop sharing the cd _table in subsequent commits.
--- arm_smmu_write_ctx_desc is used to sync the cd for all masters the domain was attached to. Now that SVA is calling it in a loop, and to break the dependency on arm_smmu_domain, we should only sync for the master passed in as the parameter
5. Introduce a skip_cd_sync parameter to arm_smmu_write_ctx_desc so that arm_smmu_domain_finalise_s1 can skip the sync_cd before the cd_table is attached to the master. Before the previous commit, arm_smmu_domain_finalise_s1 was calling arm_smmu_write_ctx_desc() before adding the master to the domain->devices list, which was used by arm_sync_smmu() to issue sync commands to masters. This optimization was broken by the previous commit since we stopped depending on domain->devices.
6. Move cd_table from domain to arm_smmu_master->cd_table.
Re: [PATCH v4 02/13] iommu/arm-smmu-v3: Add smmu_s1_cfg to smmu_master
Quoted 10 lines · Michael Shavit
Awesome that helps a lot. These are the patches I have in mind:
1. Move
ctx_descout ofs1_cfg
2. Replacedomain->s1_cfgwithstruct arm_smmu_ctx_desc_cfg cd_table
3. Movestall_enabledfrom domain toarm_smmu_ctx_desc_cfg
4. Accept anarm_smmu_masterinstead of anarm_smmu_domainas the parameter forarm_smmu_write_ctx_desc
---arm_smmu_write_ctx_desccan simply get thecd_tableit writes to frommaster->domain->cd_table; this get's replaced bymaster->cd_tablein subsequent commits.
Makes sense
Quoted 5 lines
--- SVA is weird if we cut changes here: it iterates over all masters a domain is attached to in order to call
arm_smmu_write_ctx_desc(), which ends up writing to the sharedcd_tablesince inmaster->domain. This becomes more sensible once masters stop sharing the cd_tablein subsequent commits.
Seems like it is OK inside the series
Quoted 4 lines
---
arm_smmu_write_ctx_descis used to sync the cd for all masters the domain was attached to. Now that SVA is calling it in a loop, and to break the dependency onarm_smmu_domain, we should only sync for the master passed in as the parameter
Sounds correct
Quoted 8 lines
5. Introduce a
skip_cd_syncparameter toarm_smmu_write_ctx_descso thatarm_smmu_domain_finalise_s1can skip thesync_cdbefore thecd_tableis attached to the master. Before the previous commit,arm_smmu_domain_finalise_s1was callingarm_smmu_write_ctx_desc()before adding the master to thedomain->deviceslist, which was used byarm_sync_smmu()to issue sync commands to masters. This optimization was broken by the previous commit since we stopped depending ondomain->devices.
Tracking if the cd table is not yet installed in a STE might be a cleaner approach than a flag?
Quoted 1 line
6. Move
cd_tablefrom domain toarm_smmu_master->cd_table.
Yep
Jason
PATCH v4 3/13 iommu/arm-smmu-v3: Refactor write_strtab_ent
In reply to Michael ShavitExplicity keep track of the s1_cfg and s2_cfg that are attached to a master in arm_smmu_master, regardless of whether they are owned by arm_smmu_master, arm_smmu_domain or userspace.
--- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 37 +++++++++------------ drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h | 2 ++ 2 files changed, 17 insertions(+), 22 deletions(-)
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +15 -22
@@ -1269,10 +1269,9 @@ static void arm_smmu_write_strtab_ent(struct arm_smmu_master *master, u32 sid,*/ u64 val =le64_to_cpu (dst[0 ]);bool ste_live = false; -struct arm_smmu_device *smmu =NULL ; +struct arm_smmu_device *smmu = master->smmu;struct arm_smmu_s1_cfg *s1_cfg =NULL ;struct arm_smmu_s2_cfg *s2_cfg =NULL ; -struct arm_smmu_domain *smmu_domain =NULL ;struct arm_smmu_cmdq_ent prefetch_cmd = { .opcode =CMDQ_OP_PREFETCH_CFG , .prefetch = {
@@ -1280,24 +1279,10 @@ static void arm_smmu_write_strtab_ent(struct arm_smmu_master *master, u32 sid, }, }; -if (master) { - smmu_domain = master->domain; - smmu = master->smmu; - } - -if (smmu_domain) { -switch (smmu_domain->stage) { -case ARM_SMMU_DOMAIN_S1 : - s1_cfg = &smmu_domain->s1_cfg; -break ; -case ARM_SMMU_DOMAIN_S2 : -case ARM_SMMU_DOMAIN_NESTED : - s2_cfg = &smmu_domain->s2_cfg; -break ; -default : -break ; - } - } +if (master->s1_cfg) + s1_cfg = master->s1_cfg; +else if (master->s2_cfg) + s2_cfg = master->s2_cfg;if (val &STRTAB_STE_0_V ) {switch (FIELD_GET (STRTAB_STE_0_CFG , val)) {
@@ -1319,8 +1304,8 @@ static void arm_smmu_write_strtab_ent(struct arm_smmu_master *master, u32 sid, val =STRTAB_STE_0_V ;/* Bypass/fault */ -if (!smmu_domain || !(s1_cfg || s2_cfg)) { -if (!smmu_domain && disable_bypass) +if (!(s1_cfg || s2_cfg)) { +if (disable_bypass) val |=FIELD_PREP (STRTAB_STE_0_CFG ,STRTAB_STE_0_CFG_ABORT );else val |=FIELD_PREP (STRTAB_STE_0_CFG ,STRTAB_STE_0_CFG_BYPASS );
@@ -2401,6 +2386,8 @@ static void arm_smmu_detach_dev(struct arm_smmu_master *master) master->domain =NULL ; master->ats_enabled = false; + master->s1_cfg =NULL ; + master->s2_cfg =NULL ;arm_smmu_install_ste_for_dev (master); }
@@ -2454,6 +2441,12 @@ static int arm_smmu_attach_dev(struct iommu_domain *domain, struct device *dev) } master->domain = smmu_domain; +if (smmu_domain->stage ==ARM_SMMU_DOMAIN_S1 ) { + master->s1_cfg = &smmu_domain->s1_cfg; + }else if (smmu_domain->stage ==ARM_SMMU_DOMAIN_S2 || + smmu_domain->stage ==ARM_SMMU_DOMAIN_NESTED ) { + master->s2_cfg = &smmu_domain->s2_cfg; + }/* * The SMMU does not support enabling ATS with bypass. When the STE is
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h +2 -0
@@ -689,6 +689,8 @@ struct arm_smmu_master {struct list_head domain_head;struct arm_smmu_stream *streams;struct arm_smmu_s1_cfg owned_s1_cfg; +struct arm_smmu_s1_cfg *s1_cfg; +struct arm_smmu_s2_cfg *s2_cfg;unsigned int num_streams;bool ats_enabled;bool stall_enabled;
Signature · 1 line
2.41.0.162.gfafddb0af9-goog
Re: [PATCH v4 03/13] iommu/arm-smmu-v3: Refactor write_strtab_ent
Quoted 3 lines · Michael Shavit
Explicity keep track of the
s1_cfgands2_cfgthat are attached to a master inarm_smmu_master, regardless of whether they are owned byarm_smmu_master,arm_smmu_domainor userspace.
An s1_cfg is in a master while an s2_cfg is in a domain. So we we know where they are. I kinda get that having these two ptrs could ease the cleanup, especially in arm_smmu_write_strtab_ent.
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h struct arm_smmu_master { · line 693
Quoted 10 lines
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h +2
-0@@ -689,6 +689,8 @@ struct arm_smmu_master {struct list_head domain_head;struct arm_smmu_stream *streams;struct arm_smmu_s1_cfg owned_s1_cfg; +struct arm_smmu_s1_cfg *s1_cfg; +struct arm_smmu_s2_cfg *s2_cfg;
Yet, this looks a bit overcomplicated to me by having an s1_cfg that points to master->owned_s1_cfg?
I am wondering if it'd be neater to have a new struct for STE, replacing the owned_s1_cfg, s1_cfg and s2_cfg above. It could be something like struct arm_smmu_cmdq_ent, which contains all the STE fields so that in arm_smmu_write_strtab_ent we can just plainly copy to the dst[]. Also, whenever a device attaches to a domain having the necessary info needed by the STE, we update the STE struct owned by the master.
Thanks
Nic
PATCH v4 4/13 iommu/arm-smmu-v3: Refactor write_ctx_desc
In reply to Michael ShavitUpdate arm_smmu_write_ctx_desc and downstream functions to be agnostic of the CD table's ownership. Note that in practice, it will only be called in cases where the table is owned by arm_smmu_master. Whether arm_smmu_write_ctx_desc will trigger a sync is also made part of the API.
Note that this change isn't a nop refactor since SVA will call arm_smmu_write_ctx_desc in a loop for every master the domain is attached to despite the fact that they all share the same CD table. Note that the next commit ceases this sharing of the s1_cfg which makes that loop necessary.
--- .../iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c | 36 ++++++++- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 80 +++++++++++-------- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h | 6 +- 3 files changed, 81 insertions(+), 41 deletions(-)
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c +32 -4
@@ -45,10 +45,12 @@ static struct arm_smmu_ctx_desc * arm_smmu_share_asid(struct mm_struct *mm, u16 asid) { int ret; +unsigned long flags;u32 new_asid;struct arm_smmu_ctx_desc *cd;struct arm_smmu_device *smmu;struct arm_smmu_domain *smmu_domain; +struct arm_smmu_master *master; cd =xa_load (&arm_smmu_asid_xa, asid);if (!cd)
@@ -80,7 +82,11 @@ arm_smmu_share_asid(struct mm_struct *mm, u16 asid)* be some overlap between use of both ASIDs, until we invalidate the * TLB. */ -arm_smmu_write_ctx_desc (smmu_domain,0 , cd); +spin_lock_irqsave (&smmu_domain->devices_lock, flags); +list_for_each_entry (master, &smmu_domain->devices, domain_head) { +arm_smmu_write_ctx_desc (smmu, master->s1_cfg, master,0 , cd); + } +spin_unlock_irqrestore (&smmu_domain->devices_lock, flags);/* Invalidate TLB entries previously associated with that context */ arm_smmu_tlb_inv_asid (smmu, asid);
@@ -211,6 +217,8 @@ static void arm_smmu_mm_release(struct mmu_notifier *mn, struct mm_struct *mm) {struct arm_smmu_mmu_notifier *smmu_mn =mn_to_smmu (mn);struct arm_smmu_domain *smmu_domain = smmu_mn->domain; +struct arm_smmu_master *master; +unsigned long flags;mutex_lock (&sva_lock);if (smmu_mn->cleared) {
@@ -222,7 +230,12 @@ static void arm_smmu_mm_release(struct mmu_notifier *mn, struct mm_struct *mm)* DMA may still be running. Keep the cd valid to avoid C_BAD_CD events, * but disable translation. */ -arm_smmu_write_ctx_desc (smmu_domain, mm->pasid, &quiet_cd); +spin_lock_irqsave (&smmu_domain->devices_lock, flags); +list_for_each_entry (master, &smmu_domain->devices, domain_head) { +arm_smmu_write_ctx_desc (master->smmu, master->s1_cfg, master, + mm->pasid, &quiet_cd); + } +spin_unlock_irqrestore (&smmu_domain->devices_lock, flags);arm_smmu_tlb_inv_asid (smmu_domain->smmu, smmu_mn->cd->asid);arm_smmu_atc_inv_domain (smmu_domain, mm->pasid,0 ,0 );
@@ -248,8 +261,10 @@ arm_smmu_mmu_notifier_get(struct arm_smmu_domain *smmu_domain, struct mm_struct *mm) {int ret; +unsigned long flags;struct arm_smmu_ctx_desc *cd;struct arm_smmu_mmu_notifier *smmu_mn; +struct arm_smmu_master *master;list_for_each_entry (smmu_mn, &smmu_domain->mmu_notifiers, list) {if (smmu_mn->mn.mm == mm) {
@@ -279,7 +294,12 @@ arm_smmu_mmu_notifier_get(struct arm_smmu_domain *smmu_domain, goto err_free_cd; } - ret =arm_smmu_write_ctx_desc (smmu_domain, mm->pasid, cd); +spin_lock_irqsave (&smmu_domain->devices_lock, flags); +list_for_each_entry (master, &smmu_domain->devices, domain_head) { + ret =arm_smmu_write_ctx_desc (master->smmu, master->s1_cfg, + master, mm->pasid, cd); + } +spin_unlock_irqrestore (&smmu_domain->devices_lock, flags);if (ret) goto err_put_notifier;
@@ -296,15 +316,23 @@ arm_smmu_mmu_notifier_get(struct arm_smmu_domain *smmu_domain,static void arm_smmu_mmu_notifier_put (struct arm_smmu_mmu_notifier *smmu_mn) { +unsigned long flags;struct mm_struct *mm = smmu_mn->mn.mm;struct arm_smmu_ctx_desc *cd = smmu_mn->cd; +struct arm_smmu_master *master;struct arm_smmu_domain *smmu_domain = smmu_mn->domain;if (!refcount_dec_and_test (&smmu_mn->refs))return ;list_del (&smmu_mn->list); -arm_smmu_write_ctx_desc (smmu_domain, mm->pasid,NULL ); + +spin_lock_irqsave (&smmu_domain->devices_lock, flags); +list_for_each_entry (master, &smmu_domain->devices, domain_head) { +arm_smmu_write_ctx_desc (master->smmu, master->s1_cfg, master, + mm->pasid,NULL ); + } +spin_unlock_irqrestore (&smmu_domain->devices_lock, flags);/* * If we went through clear(), we've already invalidated, and no
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +45 -35
178 lines not on this page · raw message
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h +4 -2
@@ -595,6 +595,7 @@ struct arm_smmu_s1_cfg {struct arm_smmu_ctx_desc_cfg cdcfg;u8 s1fmt;u8 s1cdmax; +bool stall_enabled; };struct arm_smmu_s2_cfg {
@@ -713,7 +714,6 @@ struct arm_smmu_domain {struct mutex init_mutex;/* Protects smmu pointer */ struct io_pgtable_ops *pgtbl_ops; -bool stall_enabled;atomic_t nr_ats_masters;enum arm_smmu_domain_stage stage;
@@ -742,7 +742,9 @@ extern struct xarray arm_smmu_asid_xa;extern struct mutex arm_smmu_asid_lock;extern struct arm_smmu_ctx_desc quiet_cd; -int arm_smmu_write_ctx_desc (struct arm_smmu_domain *smmu_domain,int ssid, +int arm_smmu_write_ctx_desc (struct arm_smmu_device *smmu, +struct arm_smmu_s1_cfg *s1_cfg, +struct arm_smmu_master *smmu_master,int ssid,struct arm_smmu_ctx_desc *cd);void arm_smmu_tlb_inv_asid (struct arm_smmu_device *smmu,u16 asid); void arm_smmu_tlb_inv_range_asid(unsigned long iova, size_t size, int asid,
Signature · 1 line
2.41.0.162.gfafddb0af9-goog
PATCH v4 5/13 iommu/arm-smmu-v3: Use the master-owned s1_cfg
In reply to Michael ShavitInsert CDs for STAGE_1 domains into a CD table owned by the arm_smmu_master. Remove the CD table that was owned by arm_smmu_domain.
--- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 43 ++++++--------------- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h | 3 -- 2 files changed, 12 insertions(+), 34 deletions(-)
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +12 -31
@@ -2065,12 +2065,8 @@ static void arm_smmu_domain_free(struct iommu_domain *domain)/* Free the CD and ASID, if we allocated them */ if (smmu_domain->stage ==ARM_SMMU_DOMAIN_S1 ) { -struct arm_smmu_s1_cfg *cfg = &smmu_domain->s1_cfg; -/* Prevent SVA from touching the CD while we're freeing it */ mutex_lock (&arm_smmu_asid_lock); -if (cfg->cdcfg.cdtab) -arm_smmu_free_cd_tables (smmu, &cfg->cdcfg);arm_smmu_free_asid (&smmu_domain->cd);mutex_unlock (&arm_smmu_asid_lock); }else {
@@ -2082,14 +2078,13 @@ static void arm_smmu_domain_free(struct iommu_domain *domain)kfree (smmu_domain); } -static int arm_smmu_domain_finalise_s1 (struct arm_smmu_domain *smmu_domain, +static int arm_smmu_domain_finalise_cd (struct arm_smmu_domain *smmu_domain,struct arm_smmu_master *master,struct io_pgtable_cfg *pgtbl_cfg) {int ret;u32 asid;struct arm_smmu_device *smmu = smmu_domain->smmu; -struct arm_smmu_s1_cfg *cfg = &smmu_domain->s1_cfg;struct arm_smmu_ctx_desc *cd = &smmu_domain->cd; typeof(&pgtbl_cfg->arm_lpae_s1_cfg.tcr) tcr = &pgtbl_cfg->arm_lpae_s1_cfg.tcr;
@@ -2102,10 +2097,6 @@ static int arm_smmu_domain_finalise_s1(struct arm_smmu_domain *smmu_domain,if (ret) goto out_unlock; - ret =arm_smmu_init_s1_cfg (master, cfg); -if (ret) - goto out_free_asid; - cd->asid = (u16 )asid; cd->ttbr = pgtbl_cfg->arm_lpae_s1_cfg.ttbr; cd->tcr = FIELD_PREP(CTXDESC_CD_0_TCR_T0SZ, tcr->tsz) |
@@ -2117,19 +2108,9 @@ static int arm_smmu_domain_finalise_s1(struct arm_smmu_domain *smmu_domain,CTXDESC_CD_0_TCR_EPD1 |CTXDESC_CD_0_AA64 ; cd->mair = pgtbl_cfg->arm_lpae_s1_cfg.mair; - ret =arm_smmu_write_ctx_desc (smmu, cfg, -NULL /*Not attached to a master yet */ , -0 , cd); -if (ret) - goto out_free_cd_tables; -mutex_unlock (&arm_smmu_asid_lock);return 0 ; -out_free_cd_tables: -arm_smmu_free_cd_tables (smmu, &cfg->cdcfg); -out_free_asid: -arm_smmu_free_asid (cd); out_unlock:mutex_unlock (&arm_smmu_asid_lock);return ret;
@@ -2192,7 +2173,7 @@ static int arm_smmu_domain_finalise(struct iommu_domain *domain, ias = min_t(unsigned long, ias, VA_BITS); oas = smmu->ias; fmt =ARM_64_LPAE_S1 ; - finalise_stage_fn = arm_smmu_domain_finalise_s1; + finalise_stage_fn = arm_smmu_domain_finalise_cd;break ;case ARM_SMMU_DOMAIN_NESTED :case ARM_SMMU_DOMAIN_S2 :
@@ -2439,20 +2420,20 @@ static int arm_smmu_attach_dev(struct iommu_domain *domain, struct device *dev) }else if (smmu_domain->smmu != smmu) { ret = -EINVAL ; goto out_unlock; - }else if (smmu_domain->stage ==ARM_SMMU_DOMAIN_S1 && - master->ssid_bits != smmu_domain->s1_cfg.s1cdmax) { - ret = -EINVAL ; - goto out_unlock; - }else if (smmu_domain->stage ==ARM_SMMU_DOMAIN_S1 && - smmu_domain->s1_cfg.stall_enabled != - master->stall_enabled) { - ret = -EINVAL ; - goto out_unlock; } master->domain = smmu_domain;if (smmu_domain->stage ==ARM_SMMU_DOMAIN_S1 ) { - master->s1_cfg = &smmu_domain->s1_cfg; + master->s1_cfg = &master->owned_s1_cfg; + ret =arm_smmu_write_ctx_desc ( + smmu, + master->s1_cfg,NULL /*Not attached to a master yet */ , +0 , &smmu_domain->cd); +if (ret) { + master->s1_cfg =NULL ; + master->domain =NULL ; + goto out_unlock; + } }else if (smmu_domain->stage ==ARM_SMMU_DOMAIN_S2 || smmu_domain->stage ==ARM_SMMU_DOMAIN_NESTED ) { master->s2_cfg = &smmu_domain->s2_cfg;
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h +0 -3
@@ -718,10 +718,7 @@ struct arm_smmu_domain {enum arm_smmu_domain_stage stage;union { -struct {struct arm_smmu_ctx_desc cd; -struct arm_smmu_s1_cfg s1_cfg; - };struct arm_smmu_s2_cfg s2_cfg; };
Signature · 1 line
2.41.0.162.gfafddb0af9-goog
Re: [PATCH v4 05/13] iommu/arm-smmu-v3: Use the master-owned s1_cfg
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h struct arm_smmu_domain { · line 723
Quoted 13 lines · Michael Shavit
Insert CDs for
STAGE_1domains into a CD table owned by thearm_smmu_master. Remove the CD table that was owned byarm_smmu_domain.drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h +0
-3@@ -718,10 +718,7 @@ struct arm_smmu_domain {enum arm_smmu_domain_stage stage;union { -struct {struct arm_smmu_ctx_desc cd; -struct arm_smmu_s1_cfg s1_cfg; - };struct arm_smmu_s2_cfg s2_cfg; };
So the arm_smmu_domain looks like an object representing either:
1) a CD table + (if !IDENTITY) an S1 IOPT for default substream
2) an S2 IOPT
I wonder if we need something like struct arm_smmu_subdomain to represent a substream, because now an IOMMU_DOMAIN_SVA domain is represented by this struct arm_smmu_domain too, which is neither of the two cases in the list above.
Thanks
Nicolin
Re: [PATCH v4 05/13] iommu/arm-smmu-v3: Use the master-owned s1_cfg
Quoted 7 lines · Nicolin Chen
Quoted 5 lines · Michael Shavit
Insert CDs for
STAGE_1domains into a CD table owned by thearm_smmu_master. Remove the CD table that was owned byarm_smmu_domain.(quoted diff) +0
-0@@ -718,10 +718,7 @@ struct arm_smmu_domain {
Quoted 12 lines
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h struct arm_smmu_domain { · line 723
Quoted 8 lines
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h +0
-3enum arm_smmu_domain_stage stage;union { -struct {struct arm_smmu_ctx_desc cd; -struct arm_smmu_s1_cfg s1_cfg; - };struct arm_smmu_s2_cfg s2_cfg; };So the
arm_smmu_domainlooks like an object representing either:
1) a CD table + (if !IDENTITY) an S1 IOPT for default substream
2) an S2 IOPT
["]
Quoted 4 lines
I wonder if we need something like struct
arm_smmu_subdomainto represent a substream, because now anIOMMU_DOMAIN_SVAdomain is represented by this structarm_smmu_domaintoo, which is neither of the two cases in the list above.
["]
I got this part incorrect: an SVA domain is actually represented by an arm_smmu_sva_domain, so there's no nested arm_smmu_domain situation. And we use arm_smmu_domain for other purpose, where I got confused.
Please ignore this part, though I am still wondering if we could name it in a way having a clear topology.
Thanks
Nicolin
PATCH v4 6/13 iommu/arm-smmu-v3: Simplify arm_smmu_enable_ats
In reply to Michael Shavitarm_smmu_enable_ats's call to inv_domain would trigger an invalidation for all masters that a domain is attached to everytime it's attached to another ATS-enabled master. It doesn't seem like those invalidations are necessary, and it's easier to reason about arm_smmu_enable_ats if it only issues invalidation commands for the current master.
---v1->v2: Fix commit message wrapping
--- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +1 -1
@@ -2286,7 +2286,7 @@ static void arm_smmu_enable_ats(struct arm_smmu_master *master) pdev =to_pci_dev (master->dev);atomic_inc (&smmu_domain->nr_ats_masters); -arm_smmu_atc_inv_domain (smmu_domain,0 ,0 ,0 ); +arm_smmu_atc_inv_master (master);if (pci_enable_ats (pdev, stu))dev_err (master->dev,"Failed to enable ATS (STU %zu)\n" , stu); }
Signature · 1 line
2.41.0.162.gfafddb0af9-goog
PATCH v4 7/13 iommu/arm-smmu-v3: Keep track of attached ssids
In reply to Michael ShavitThe arm-smmu-v3 driver keeps track of all masters that a domain is attached to so that it can re-write their STEs when the domain's ASID is upated by SVA. This tracking is also used to invalidate ATCs on all masters that a domain is attached to.
This change introduces a new data structures to track all the CD entries that a domain is attached to. This change is a pre-requisite to allow domain attachment on non 0 SSIDs.
---v3->v4: Remove reference to the master's domain accidentally re-introduced
during a rebase.
Make arm_smmu_atc_inv_domain static.v1->v2: Fix arm_smmu_atc_inv_cmd_set_ssid and other cosmetic changes
--- .../iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c | 53 +++++++---- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 89 ++++++++++++------- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h | 18 ++-- 3 files changed, 105 insertions(+), 55 deletions(-)
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c +35 -18
@@ -51,6 +51,7 @@ arm_smmu_share_asid(struct mm_struct *mm, u16 asid)struct arm_smmu_device *smmu;struct arm_smmu_domain *smmu_domain;struct arm_smmu_master *master; +struct arm_smmu_attached_domain *attached_domain; cd =xa_load (&arm_smmu_asid_xa, asid);if (!cd)
@@ -82,11 +83,14 @@ arm_smmu_share_asid(struct mm_struct *mm, u16 asid)* be some overlap between use of both ASIDs, until we invalidate the * TLB. */ -spin_lock_irqsave (&smmu_domain->devices_lock, flags); -list_for_each_entry (master, &smmu_domain->devices, domain_head) { -arm_smmu_write_ctx_desc (smmu, master->s1_cfg, master,0 , cd); +spin_lock_irqsave (&smmu_domain->attached_domains_lock, flags); +list_for_each_entry (attached_domain, &smmu_domain->attached_domains, + domain_head) { + master = attached_domain->master; +arm_smmu_write_ctx_desc (smmu, master->s1_cfg, master, + attached_domain->ssid, cd); } -spin_unlock_irqrestore (&smmu_domain->devices_lock, flags); +spin_unlock_irqrestore (&smmu_domain->attached_domains_lock, flags);/* Invalidate TLB entries previously associated with that context */ arm_smmu_tlb_inv_asid (smmu, asid);
@@ -210,7 +214,7 @@ static void arm_smmu_mm_invalidate_range(struct mmu_notifier *mn,if (!(smmu_domain->smmu->features &ARM_SMMU_FEAT_BTM ))arm_smmu_tlb_inv_range_asid (start, size, smmu_mn->cd->asid,PAGE_SIZE , false, smmu_domain); -arm_smmu_atc_inv_domain (smmu_domain, mm->pasid, start, size); +arm_smmu_atc_inv_domain_ssid (smmu_domain, mm->pasid, start, size); }static void arm_smmu_mm_release (struct mmu_notifier *mn,struct mm_struct *mm)
@@ -218,6 +222,7 @@ static void arm_smmu_mm_release(struct mmu_notifier *mn, struct mm_struct *mm)struct arm_smmu_mmu_notifier *smmu_mn =mn_to_smmu (mn);struct arm_smmu_domain *smmu_domain = smmu_mn->domain;struct arm_smmu_master *master; +struct arm_smmu_attached_domain *attached_domain;unsigned long flags;mutex_lock (&sva_lock);
@@ -230,15 +235,21 @@ static void arm_smmu_mm_release(struct mmu_notifier *mn, struct mm_struct *mm)* DMA may still be running. Keep the cd valid to avoid C_BAD_CD events, * but disable translation. */ -spin_lock_irqsave (&smmu_domain->devices_lock, flags); -list_for_each_entry (master, &smmu_domain->devices, domain_head) { -arm_smmu_write_ctx_desc (master->smmu, master->s1_cfg, master, - mm->pasid, &quiet_cd); +spin_lock_irqsave (&smmu_domain->attached_domains_lock, flags); +list_for_each_entry (attached_domain, &smmu_domain->attached_domains, + domain_head) { + master = attached_domain->master; +/* +* SVA domains piggyback on the attached_domain with SSID 0. +*/ +if (attached_domain->ssid ==0 ) +arm_smmu_write_ctx_desc (master->smmu, master->s1_cfg, + master, mm->pasid, &quiet_cd); } -spin_unlock_irqrestore (&smmu_domain->devices_lock, flags); +spin_unlock_irqrestore (&smmu_domain->attached_domains_lock, flags);arm_smmu_tlb_inv_asid (smmu_domain->smmu, smmu_mn->cd->asid); -arm_smmu_atc_inv_domain (smmu_domain, mm->pasid,0 ,0 ); +arm_smmu_atc_inv_domain_ssid (smmu_domain, mm->pasid,0 ,0 ); smmu_mn->cleared = true;mutex_unlock (&sva_lock);
@@ -265,6 +276,7 @@ arm_smmu_mmu_notifier_get(struct arm_smmu_domain *smmu_domain,struct arm_smmu_ctx_desc *cd;struct arm_smmu_mmu_notifier *smmu_mn;struct arm_smmu_master *master; +struct arm_smmu_attached_domain *attached_domain;list_for_each_entry (smmu_mn, &smmu_domain->mmu_notifiers, list) {if (smmu_mn->mn.mm == mm) {
@@ -294,12 +306,14 @@ arm_smmu_mmu_notifier_get(struct arm_smmu_domain *smmu_domain, goto err_free_cd; } -spin_lock_irqsave (&smmu_domain->devices_lock, flags); -list_for_each_entry (master, &smmu_domain->devices, domain_head) { +spin_lock_irqsave (&smmu_domain->attached_domains_lock, flags); +list_for_each_entry (attached_domain, &smmu_domain->attached_domains, + domain_head) { + master = attached_domain->master; ret =arm_smmu_write_ctx_desc (master->smmu, master->s1_cfg, master, mm->pasid, cd); } -spin_unlock_irqrestore (&smmu_domain->devices_lock, flags); +spin_unlock_irqrestore (&smmu_domain->attached_domains_lock, flags);if (ret) goto err_put_notifier;
@@ -319,6 +333,7 @@ static void arm_smmu_mmu_notifier_put(struct arm_smmu_mmu_notifier *smmu_mn)unsigned long flags;struct mm_struct *mm = smmu_mn->mn.mm;struct arm_smmu_ctx_desc *cd = smmu_mn->cd; +struct arm_smmu_attached_domain *attached_domain;struct arm_smmu_master *master;struct arm_smmu_domain *smmu_domain = smmu_mn->domain;
@@ -327,12 +342,14 @@ static void arm_smmu_mmu_notifier_put(struct arm_smmu_mmu_notifier *smmu_mn)list_del (&smmu_mn->list); -spin_lock_irqsave (&smmu_domain->devices_lock, flags); -list_for_each_entry (master, &smmu_domain->devices, domain_head) { +spin_lock_irqsave (&smmu_domain->attached_domains_lock, flags); +list_for_each_entry (attached_domain, &smmu_domain->attached_domains, + domain_head) { + master = attached_domain->master;arm_smmu_write_ctx_desc (master->smmu, master->s1_cfg, master, mm->pasid,NULL ); } -spin_unlock_irqrestore (&smmu_domain->devices_lock, flags); +spin_unlock_irqrestore (&smmu_domain->attached_domains_lock, flags);/* * If we went through clear(), we've already invalidated, and no
@@ -340,7 +357,7 @@ static void arm_smmu_mmu_notifier_put(struct arm_smmu_mmu_notifier *smmu_mn)*/ if (!smmu_mn->cleared) {arm_smmu_tlb_inv_asid (smmu_domain->smmu, cd->asid); -arm_smmu_atc_inv_domain (smmu_domain, mm->pasid,0 ,0 ); +arm_smmu_atc_inv_domain_ssid (smmu_domain, mm->pasid,0 ,0 ); }/* Frees smmu_mn */
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +57 -32
213 lines not on this page · raw message
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h +13 -5
@@ -682,11 +682,19 @@ struct arm_smmu_stream {struct rb_node node; }; +/* List of {masters, ssid} that a domain is attached to */ +struct arm_smmu_attached_domain { +struct list_head domain_head; +struct arm_smmu_domain *domain; +struct arm_smmu_master *master; +int ssid; +}; +/* SMMU private data for each master */ struct arm_smmu_master {struct arm_smmu_device *smmu;struct device *dev; -struct arm_smmu_domain *domain; +struct arm_smmu_attached_domain non_pasid_domain;struct list_head domain_head;struct arm_smmu_stream *streams;struct arm_smmu_s1_cfg owned_s1_cfg;
@@ -724,8 +732,8 @@ struct arm_smmu_domain {struct iommu_domain domain; -struct list_head devices; -spinlock_t devices_lock; +struct list_head attached_domains; +spinlock_t attached_domains_lock;struct list_head mmu_notifiers; };
@@ -748,8 +756,8 @@ void arm_smmu_tlb_inv_range_asid(unsigned long iova, size_t size, int asid, size_t granule, bool leaf, struct arm_smmu_domain *smmu_domain);bool arm_smmu_free_asid (struct arm_smmu_ctx_desc *cd); -int arm_smmu_atc_inv_domain (struct arm_smmu_domain *smmu_domain,int ssid, -unsigned long iova,size_t size); +int arm_smmu_atc_inv_domain_ssid (struct arm_smmu_domain *smmu_domain,int ssid, +unsigned long iova,size_t size);#ifdef CONFIG_ARM_SMMU_V3_SVA bool arm_smmu_sva_supported (struct arm_smmu_device *smmu);
Signature · 1 line
2.41.0.162.gfafddb0af9-goog
Re: [PATCH v4 07/13] iommu/arm-smmu-v3: Keep track of attached ssids
Quoted 18 lines · Michael Shavit
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h +9
-1+/* List of {masters, ssid} that a domain is attached to */ +struct arm_smmu_attached_domain { +struct list_head domain_head; +struct arm_smmu_domain *domain; +struct arm_smmu_master *master; +int ssid; +}; +/* SMMU private data for each master */ struct arm_smmu_master {struct arm_smmu_device *smmu;struct device *dev; -struct arm_smmu_domain *domain; +struct arm_smmu_attached_domain non_pasid_domain;struct list_head domain_head;struct arm_smmu_stream *streams;struct arm_smmu_s1_cfg owned_s1_cfg;
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h struct arm_smmu_domain { · line 736
Quoted 7 lines
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h +2
-2@@ -724,8 +732,8 @@ struct arm_smmu_domain {struct iommu_domain domain; -struct list_head devices; -spinlock_t devices_lock; +struct list_head attached_domains; +spinlock_t attached_domains_lock;
Yea, I think that this arm_smmu_attached_domain is similar to the "subdomain" that I was talking about in the previous reply, though having a list of attached domains under a domain doesn't feel very clear.
Perhaps it would be good to have some renaming and kdoc too.
And since we have a group of subdomains that are simply indexed by ssids, perhaps we can add an xarray to store a subdomain ptr along with an ssid, replacing the list?
Thanks
Nicolin
Re: [PATCH v4 07/13] iommu/arm-smmu-v3: Keep track of attached ssids
Quoted 3 lines
And since we have a group of subdomains that are simply indexed by ssids, perhaps we can add an xarray to store a subdomain ptr along with an ssid, replacing the list?
Hmmmm, I think the only place where we search through the list for a specific SSID is during remove_dev_pasid. We mostly use the list to iterate over all the masters/ssids that this domain is attached to. I'm not sure if moving to an xarray to optimize the remove_dev_pasid call is worth it (at the cost of iterations which IIUC would become N log(N))
Re: [PATCH v4 07/13] iommu/arm-smmu-v3: Keep track of attached ssids
Quoted 11 lines · Michael Shavit
Quoted 4 lines
And since we have a group of subdomains that are simply indexed by ssids, perhaps we can add an xarray to store a subdomain ptr along with an ssid, replacing the list?
Hmmmm, I think the only place where we search through the list for a specific SSID is during
remove_dev_pasid. We mostly use the list to iterate over all the masters/ssids that this domain is attached to. I'm not sure if moving to an xarray to optimize theremove_dev_pasidcall is worth it (at the cost of iterations which IIUC would become N log(N))
OK. That's fine.
Can we have a smaller rework series first? Let's have all data structures defined. And then we can add features on top of it.
Thanks
Nicolin
Re: [PATCH v4 07/13] iommu/arm-smmu-v3: Keep track of attached ssids
In reply to Michael ShavitQuoted 9 lines · Michael Shavit
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +7
-2-int arm_smmu_atc_inv_domain (struct arm_smmu_domain *smmu_domain,int ssid, -unsigned long iova,size_t size) +/* +* If ssid is non-zero, issue atc invalidations with the given ssid instead of +* the one the domain is attached to. This is used by SVA since it's pasid +* attachments aren't recorded in smmu_domain yet. +*/ +int arm_smmu_atc_inv_domain_ssid (struct arm_smmu_domain *smmu_domain,int ssid, +unsigned long iova,size_t size)
[..]
Quoted 17 lines
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +5
-3@@ -1823,25 +1835,37 @@ int arm_smmu_atc_inv_domain(struct arm_smmu_domain *smmu_domain, int ssid,if (!atomic_read (&smmu_domain->nr_ats_masters))return 0 ; -arm_smmu_atc_inv_to_cmd (ssid, iova, size, &cmd); +arm_smmu_atc_inv_to_cmd (iova, size, &cmd); cmds.num =0 ; -spin_lock_irqsave (&smmu_domain->devices_lock, flags); -list_for_each_entry (master, &smmu_domain->devices, domain_head) { +spin_lock_irqsave (&smmu_domain->attached_domains_lock, flags); +list_for_each_entry (attached_domain, &smmu_domain->attached_domains, + domain_head) { + master = attached_domain->master;if (!master->ats_enabled)continue ;
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c int arm_smmu_atc_inv_domain(struct arm_smmu_domain *smmu_domain, int ssid, · line 1857
Quoted 10 lines
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +4
-0+if (ssid !=0 ) +arm_smmu_atc_inv_cmd_set_ssid (ssid, &cmd); +else +arm_smmu_atc_inv_cmd_set_ssid (attached_domain->ssid, &cmd);for (i =0 ; i < master->num_streams; i++) { cmd.atc.sid = master->streams[i].id;arm_smmu_cmdq_batch_add (smmu_domain->smmu, &cmds, &cmd); } }
And I don't quite get this part. Prior to this change, it issues one ATC_INV command covering all ATC entries per comments inside arm_smmu_atc_inv_to_cmd(). But now we replace that single command with all attached subdomains in the list? Any reason for such a change here?
Thanks
Nicolin
Re: [PATCH v4 07/13] iommu/arm-smmu-v3: Keep track of attached ssids
Quoted 5 lines
And I don't quite get this part. Prior to this change, it issues one
ATC_INVcommand covering all ATC entries per comments insidearm_smmu_atc_inv_to_cmd(). But now we replace that single command with all attached subdomains in the list? Any reason for such a change here?
Because we don't necessarily want to invalidate all PASID-domains attached to a master. If arm_smmu_atc_inv_domain() is called on a domain that is only attached with Pasid, we can restrict the invalidations to those specific PASID by looping over them. But yeah, you're right that we could potentially optimize this?
* Skip the per-pasid invalidations if the domain is also attached to this master without PASID as we have to invalidate all its pasids in that case anyways. It's hard to imagine clients attaching a domain both with pasid and without pasid to the same device but could be possible.
* Always invalidate all pasids by issuing atc invalidations on SSID 0. This sounds like the wrong trade-off??
Re: [PATCH v4 07/13] iommu/arm-smmu-v3: Keep track of attached ssids
Quoted 11 lines · Michael Shavit
Quoted 5 lines
And I don't quite get this part. Prior to this change, it issues one
ATC_INVcommand covering all ATC entries per comments insidearm_smmu_atc_inv_to_cmd(). But now we replace that single command with all attached subdomains in the list? Any reason for such a change here?Because we don't necessarily want to invalidate all PASID-domains attached to a master. If
arm_smmu_atc_inv_domain()is called on a domain that is only attached with Pasid, we can restrict the invalidations to those specific PASID by looping over them.
Quoted 9 lines
But yeah,
you're right that we could potentially optimize this?
* Skip the per-pasid invalidations if the domain is also attached to this master without PASID as we have to invalidate all its pasids in that case anyways. It's hard to imagine clients attaching a domain both with pasid and without pasid to the same device but could be possible.
* Always invalidate all pasids by issuing atc invalidations on SSID 0. This sounds like the wrong trade-off??
Well, firstly it's kinda odd to have this optimization hidden in a big rework patch. And I am not sure if it alone would work for all use cases, as it impacts the arm_smmu_atc_inv_domain() that passes in a zero ssid, in which case the affected function is not used by a pasid case all the time:
/* * ATS and PASID:
...
* When using STRTAB_STE_1_S1DSS_SSID0 (reserving CD 0 for non-PASID
* traffic), translation requests without PASID create ATC entries
* without PASID, which must be invalidated with substream_valid clear.
* This has the unpleasant side-effect of invalidating all PASID-tagged
* ATC entries within the address range.
*/
Thanks
Nicolin
Re: [PATCH v4 07/13] iommu/arm-smmu-v3: Keep track of attached ssids
Quoted 5 lines · Nicolin Chen
Well, firstly it's kinda odd to have this optimization hidden in a big rework patch. And I am not sure if it alone would work for all use cases, as it impacts the
arm_smmu_atc_inv_domain()that passes in a zero ssid, in which case the affected function is not used by a pasid case all the time:
To clarify, we haven't changed anything for the existing flow. Before the latter set_dev_pasid patch is introduced, there's only ever a single {master,domain} pair in the attached_domain list. When arm_smmu_atc_inv_domain() is called on a non-pasid domain, it issues a single atc inv command with substream_valid=false, same as before. When SVA wants to invalidate a domain, it calls arm_smmu_atc_inv_domain_ssid to issue a single atc inv command with that specific PASID, same as before. At no point does the driver ever rely on the fact that invalidating with substream_valid=false would invalidate all PASIDs.
This patch series simply follows the precedent set by SVA: when a domain is attached with a pasid, invalidates those specific pasids. It'd feel kinda hacky to rely on the fact that invalidating with substream_valid=false would invalidate all PASIDs.
PATCH v4 8/13 iommu/arm-smmu-v3: Add helper for atc invalidation
In reply to Michael ShavitThis will be used to invalidate ATC entries made on an SSID for a master when detaching a domain with pasid.
---v1->v2: Make use of arm_smmu_atc_inv_cmd_set_ssid
--- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-)
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +8 -1
@@ -1784,13 +1784,15 @@ arm_smmu_atc_inv_to_cmd(unsigned long iova, size_t size, cmd->atc.size = log2_span; } -static int arm_smmu_atc_inv_master (struct arm_smmu_master *master) +static int arm_smmu_atc_inv_master_ssid (struct arm_smmu_master *master, +int ssid) {int i;struct arm_smmu_cmdq_ent cmd;struct arm_smmu_cmdq_batch cmds;arm_smmu_atc_inv_to_cmd (0 ,0 , &cmd); +arm_smmu_atc_inv_cmd_set_ssid (ssid, &cmd); cmds.num =0 ;for (i =0 ; i < master->num_streams; i++) { cmd.atc.sid = master->streams[i].id;
@@ -1800,6 +1802,11 @@ static int arm_smmu_atc_inv_master(struct arm_smmu_master *master)return arm_smmu_cmdq_batch_submit (master->smmu, &cmds); } +static int arm_smmu_atc_inv_master (struct arm_smmu_master *master) +{ +return arm_smmu_atc_inv_master_ssid (master,0 ); +} +/* * If ssid is non-zero, issue atc invalidations with the given ssid instead of * the one the domain is attached to. This is used by SVA since it's pasid
Signature · 1 line
2.41.0.162.gfafddb0af9-goog
PATCH v4 9/13 iommu/arm-smmu-v3: Implement set_dev_pasid
In reply to Michael ShavitThis change enables the use of the iommu_attach_dev_pasid API for UNMANAGED domains. The primary use-case is to allow in-kernel users of the iommu API to manage domains with PASID. This change also allows for future support of pasid in the DMA api.
---v1->v2: Add missing atc invalidation when detaching with pasid
--- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 167 +++++++++++++++++--- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h | 1 + 2 files changed, 149 insertions(+), 19 deletions(-)
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +148 -19
@@ -2173,6 +2173,10 @@ static int arm_smmu_domain_finalise_s2(struct arm_smmu_domain *smmu_domain,return 0 ; } +/* +* master may be null for domain types that are finalized before being attached +* to a master. +*/ static int arm_smmu_domain_finalise (struct iommu_domain *domain,struct arm_smmu_master *master) {
@@ -2369,6 +2373,11 @@ static int arm_smmu_enable_pasid(struct arm_smmu_master *master)return 0 ; } +static bool arm_smmu_master_has_pasid_domains (struct arm_smmu_master *master) +{ +return master->nr_attached_pasid_domains >0 ; +} +static void arm_smmu_disable_pasid (struct arm_smmu_master *master) {struct pci_dev *pdev;
@@ -2411,6 +2420,28 @@ static void arm_smmu_detach_dev(struct arm_smmu_master *master)arm_smmu_install_ste_for_dev (master); } +/* +* Once attached for the first time, a domain can no longer be attached to any +* master with a distinct upstream SMMU. +*/ +static int arm_smmu_prepare_domain_for_smmu (struct arm_smmu_device *smmu, +struct arm_smmu_domain *smmu_domain) +{ +int ret =0 ; + +mutex_lock (&smmu_domain->init_mutex); +if (!smmu_domain->smmu) { + smmu_domain->smmu = smmu; + ret =arm_smmu_domain_finalise (&smmu_domain->domain,NULL ); +if (ret) + smmu_domain->smmu =NULL ; + }else if (smmu_domain->smmu != smmu) { + ret = -EINVAL ; + } +mutex_unlock (&smmu_domain->init_mutex); +return ret; +} +static int arm_smmu_attach_dev (struct iommu_domain *domain,struct device *dev) {int ret =0 ;
@@ -2426,6 +2457,10 @@ static int arm_smmu_attach_dev(struct iommu_domain *domain, struct device *dev) master =dev_iommu_priv_get (dev); smmu = master->smmu; + ret =arm_smmu_prepare_domain_for_smmu (smmu, smmu_domain); +if (ret) +return ret; +/* * Checking that SVA is disabled ensures that this device isn't bound to * any mm, and can be safely detached from its old domain. Bonds cannot
@@ -2436,22 +2471,18 @@ static int arm_smmu_attach_dev(struct iommu_domain *domain, struct device *dev)return -EBUSY ; } -arm_smmu_detach_dev (master); - -mutex_lock (&smmu_domain->init_mutex); - -if (!smmu_domain->smmu) { - smmu_domain->smmu = smmu; - ret =arm_smmu_domain_finalise (domain, master); -if (ret) { - smmu_domain->smmu =NULL ; - goto out_unlock; - } - }else if (smmu_domain->smmu != smmu) { - ret = -EINVAL ; - goto out_unlock; +/* +* Attaching a bypass or stage 2 domain would break any domains attached +* with pasid. Attaching an S1 domain should be feasible but requires +* more complicated logic to handle. +*/ +if (arm_smmu_master_has_pasid_domains (master)) { +dev_err (dev,"cannot attach - domain attached with pasid\n" ); +return -EBUSY ; } +arm_smmu_detach_dev (master); +if (smmu_domain->stage ==ARM_SMMU_DOMAIN_S1 ) { master->s1_cfg = &master->owned_s1_cfg; ret = arm_smmu_write_ctx_desc(
@@ -2460,7 +2491,7 @@ static int arm_smmu_attach_dev(struct iommu_domain *domain, struct device *dev) 0, &smmu_domain->cd);if (ret) { master->s1_cfg =NULL ; - goto out_unlock; +return ret; } }else if (smmu_domain->stage ==ARM_SMMU_DOMAIN_S2 || smmu_domain->stage ==ARM_SMMU_DOMAIN_NESTED ) {
@@ -2489,11 +2520,75 @@ static int arm_smmu_attach_dev(struct iommu_domain *domain, struct device *dev)arm_smmu_enable_ats (master, smmu_domain); -out_unlock: -mutex_unlock (&smmu_domain->init_mutex);return ret; } +static int arm_smmu_set_dev_pasid (struct iommu_domain *domain, +struct device *dev,ioasid_t pasid) +{ +int ret =0 ; +unsigned long flags; +struct iommu_fwspec *fwspec =dev_iommu_fwspec_get (dev); +struct arm_smmu_device *smmu; +struct arm_smmu_domain *smmu_domain =to_smmu_domain (domain); +struct arm_smmu_attached_domain *attached_domain; +struct arm_smmu_master *master; + +if (!fwspec) +return -ENOENT ; + + master =dev_iommu_priv_get (dev); + smmu = master->smmu; + + ret =arm_smmu_prepare_domain_for_smmu (smmu, smmu_domain); +if (ret) +return ret; + +if (pasid ==0 ) { +dev_err (dev,"pasid 0 is reserved for the device's primary domain\n" ); +return -ENODEV ; + } + +if (smmu_domain->stage !=ARM_SMMU_DOMAIN_S1 ) { +dev_err (dev,"set_dev_pasid only supports stage 1 domains\n" ); +return -EINVAL ; + } + +if (!master->s1_cfg || master->s2_cfg) +return -EBUSY ; + + attached_domain =kzalloc (sizeof (*attached_domain),GFP_KERNEL ); +if (!attached_domain) +return -ENOMEM ; + + attached_domain->master = master; + attached_domain->domain = smmu_domain; + attached_domain->ssid = pasid; + + master->nr_attached_pasid_domains +=1 ; +/* +* arm_smmu_share_asid may update the cd's asid value and write the +* ctx_desc for every attached_domains in the list. There's a potential +* race here regardless of whether we first write the ctx_desc or +* first insert into the domain's list. Grabbing the asic_lock prevents +* SVA from changing the cd's ASID while the cd is being attached. +*/ +mutex_lock (&arm_smmu_asid_lock); + ret =arm_smmu_write_ctx_desc (master->smmu, master->s1_cfg, master, + pasid, &smmu_domain->cd); +if (ret) { +mutex_unlock (&arm_smmu_asid_lock); +kfree (attached_domain); + } + +spin_lock_irqsave (&smmu_domain->attached_domains_lock, flags); +list_add (&attached_domain->domain_head, &smmu_domain->attached_domains); +spin_unlock_irqrestore (&smmu_domain->attached_domains_lock, flags); +mutex_unlock (&arm_smmu_asid_lock); + +return 0 ; +} +static int arm_smmu_map_pages (struct iommu_domain *domain,unsigned long iova,phys_addr_t paddr,size_t pgsize,size_t pgcount,int prot,gfp_t gfp,size_t *mapped)
@@ -2739,6 +2834,15 @@ static void arm_smmu_release_device(struct device *dev)if (WARN_ON (arm_smmu_master_sva_enabled (master)))iopf_queue_remove_device (master->smmu->evtq.iopf, dev); +if (WARN_ON (master->nr_attached_pasid_domains !=0 )) { +/* +* TODO: Do we need to handle this case? +* This requires a mechanism to obtain all the pasid domains +* that this master is attached to so that we can clean up the +* domain's attached_domain list. +*/ + } +arm_smmu_detach_dev (master);arm_smmu_free_cd_tables (master->smmu, &master->owned_s1_cfg.cdcfg);arm_smmu_disable_pasid (master);
@@ -2874,12 +2978,36 @@ static int arm_smmu_def_domain_type(struct device *dev)static void arm_smmu_remove_dev_pasid (struct device *dev,ioasid_t pasid) {struct iommu_domain *domain; +struct arm_smmu_master *master =dev_iommu_priv_get (dev); +struct arm_smmu_domain *smmu_domain; +struct arm_smmu_attached_domain *attached_domain; +unsigned long flags; - domain =iommu_get_domain_for_dev_pasid (dev, pasid,IOMMU_DOMAIN_SVA ); +if (!master || pasid ==0 ) +return ; + + domain =iommu_get_domain_for_dev_pasid (dev, pasid,0 );if (WARN_ON (IS_ERR (domain)) || !domain)return ; +if (domain->type ==IOMMU_DOMAIN_SVA ) +return arm_smmu_sva_remove_dev_pasid (domain, dev, pasid); -arm_smmu_sva_remove_dev_pasid (domain, dev, pasid); + smmu_domain =to_smmu_domain (domain); +mutex_lock (&arm_smmu_asid_lock); +spin_lock_irqsave (&smmu_domain->attached_domains_lock, flags); +list_for_each_entry (attached_domain, &smmu_domain->attached_domains, domain_head) { +if (attached_domain->master != master || + attached_domain->ssid != pasid) +continue ; +list_del (&attached_domain->domain_head); +break ; + } +spin_unlock_irqrestore (&smmu_domain->attached_domains_lock, flags); +arm_smmu_write_ctx_desc (master->smmu, master->s1_cfg, master, pasid, +NULL ); +arm_smmu_atc_inv_master_ssid (master, pasid); + master->nr_attached_pasid_domains -=1 ; +mutex_unlock (&arm_smmu_asid_lock); }static struct iommu_ops arm_smmu_ops = {
@@ -2899,6 +3027,7 @@ static struct iommu_ops arm_smmu_ops = { .owner = THIS_MODULE, .default_domain_ops = &(const struct iommu_domain_ops ) { .attach_dev = arm_smmu_attach_dev, + .set_dev_pasid = arm_smmu_set_dev_pasid, .map_pages = arm_smmu_map_pages, .unmap_pages = arm_smmu_unmap_pages, .flush_iotlb_all = arm_smmu_flush_iotlb_all,
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h +1 -0
@@ -707,6 +707,7 @@ struct arm_smmu_master {bool iopf_enabled;struct list_head bonds;unsigned int ssid_bits; +unsigned int nr_attached_pasid_domains; };/* SMMU private data for an IOMMU domain */
Signature · 1 line
2.41.0.162.gfafddb0af9-goog
Re: [PATCH v4 09/13] iommu/arm-smmu-v3: Implement set_dev_pasid
Hi Michael,
Quoted 5 lines · Michael Shavit
This change enables the use of the
iommu_attach_dev_pasidAPI for UNMANAGED domains. The primary use-case is to allow in-kernel users of the iommu API to manage domains with PASID. This change also allows for future support of pasid in the DMA api.
Would you please elaborate a bit more for the use case of allowing that? And which test configuration do you cover using smmute? Would you mind sharing your test commands?
I have run a sanity with this series using an SVA domain with a real master. It seems to be fine.
Thanks
Nicolin
Re: [PATCH v4 09/13] iommu/arm-smmu-v3: Implement set_dev_pasid
Quoted 1 line
Would you please elaborate a bit more for the use case of
allowing that?
In short; to support devices that use SSID to isolate different (device) contexts. Those contexts (and the memory available to them) is partially managed by the device's Kernel driver.
Quoted 1 line
And which test configuration do you cover
using smmute? Would you mind sharing your test commands?
I used the setup suggested by Jean in
https://lore.kernel.org/all/20230511195928.GA288490@myrica/ , with the following commands:
Quoted 1 line
Quoted 1 line
Quoted 1 line
# Basic test
./smmute;
./smmute -u mmap;
# Test invalid access to not-mapped address
./smmute -u mmap -f drv;
# Test invalid access after unmap
./smmute -u mmap -f write -d;
# Check smmu_mn released when killed
mount -t tracefs nodev /sys/kernel/tracing;
echo 1 > /sys/kernel/tracing/events/iommu/enable;
echo 1 > /sys/kernel/tracing/events/smmu/enable;
./smmute -u mmap -k bind;
cat /sys/kernel/tracing/trace;
<<<
This only covers existing SVA functionality. To test the functionality introduced by this patch, I used a device capable of generating DMA transactions w/ SSID and a test driver with following tests:
1. Successful dma Read/Write to buffer mapped on domain attached with pasid
2. Unsuccessful dma Read/Write to same buffer after domain detached from pasid, or after buffer unmapped from domain
3. Variations of the above with a domain attached to multiple pasids
I've been considering migrating those tests to the smmute driver if that would be valuable.
Quoted 19 lines · Nicolin Chen
Hi Michael,
Quoted 5 lines · Michael Shavit
This change enables the use of the
iommu_attach_dev_pasidAPI for UNMANAGED domains. The primary use-case is to allow in-kernel users of the iommu API to manage domains with PASID. This change also allows for future support of pasid in the DMA api.Would you please elaborate a bit more for the use case of allowing that? And which test configuration do you cover using smmute? Would you mind sharing your test commands?
I have run a sanity with this series using an SVA domain with a real master. It seems to be fine.
Thanks
Nicolin
Re: [PATCH v4 09/13] iommu/arm-smmu-v3: Implement set_dev_pasid
Quoted 6 lines · Michael Shavit
Quoted 1 line
Would you please elaborate a bit more for the use case of
allowing that?
In short; to support devices that use SSID to isolate different (device) contexts. Those contexts (and the memory available to them) is partially managed by the device's Kernel driver.
What can be a real world use case for that?
Quoted 17 lines
Quoted 1 line
And which test configuration do you cover
using smmute? Would you mind sharing your test commands?
I used the setup suggested by Jean in
https://lore.kernel.org/all/20230511195928.GA288490@myrica/ , with the following commands:Quoted 1 line
Quoted 1 line
Quoted 1 line
# Basic test
./smmute;
./smmute -u mmap;# Test invalid access to not-mapped address
./smmute -u mmap -f drv;# Test invalid access after unmap
./smmute -u mmap -f write -d;
Quoted 7 lines
# Check
smmu_mnreleased when killed
mount -t tracefs nodev/sys/kernel/tracing;
echo 1 >/sys/kernel/tracing/events/iommu/enable;
echo 1 >/sys/kernel/tracing/events/smmu/enable;
./smmute -u mmap -k bind;
cat/sys/kernel/tracing/trace;
<<<
OK. Thanks for sharing.
Quoted 10 lines
This only covers existing SVA functionality. To test the functionality introduced by this patch, I used a device capable of generating DMA transactions w/ SSID and a test driver with following tests:
1. Successful dma Read/Write to buffer mapped on domain attached with pasid
2. Unsuccessful dma Read/Write to same buffer after domain detached from pasid, or after buffer unmapped from domain
3. Variations of the above with a domain attached to multiple pasidsI've been considering migrating those tests to the smmute driver if that would be valuable.
Is this on Gerrit too?
Thanks
Nicolin
Re: [PATCH v4 09/13] iommu/arm-smmu-v3: Implement set_dev_pasid
Quoted 4 lines · Nicolin Chen
Quoted 2 lines
I've been considering migrating those tests to the smmute driver if that would be valuable.
Is this on Gerrit too?
It's not pretty but I've rewritten the tests into the smmute kernel driver. Pushed to Gerrit here:
https://linux-review.googlesource.com/id/Ibb33ba6f9c6d069324f21b9ad98e29c94e15374a
.
Re: [PATCH v4 09/13] iommu/arm-smmu-v3: Implement set_dev_pasid
In reply to Michael ShavitQuoted 4 lines · Michael Shavit
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +4
-0+mutex_lock (&arm_smmu_asid_lock); + ret =arm_smmu_write_ctx_desc (master->smmu, master->s1_cfg, master, + pasid, &smmu_domain->cd); +if (ret) {
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c static int arm_smmu_attach_dev(struct iommu_domain *domain, struct device *dev) · line 2589
Quoted 10 lines
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +10
-0+mutex_unlock (&arm_smmu_asid_lock); +kfree (attached_domain); + } + +spin_lock_irqsave (&smmu_domain->attached_domains_lock, flags); +list_add (&attached_domain->domain_head, &smmu_domain->attached_domains); +spin_unlock_irqrestore (&smmu_domain->attached_domains_lock, flags); +mutex_unlock (&arm_smmu_asid_lock); + +return 0 ;
A small bug in this patch as well: we return 0 when arm_smmu_write_ctx_desc() fails instead of the error. Will upload fix in v5.
PATCH v4 10/13 iommu/arm-smmu-v3-sva: Remove bond refcount
In reply to Michael ShavitThe iommu-sva framework checks if a bond between a device and mm already exists and handles refcounting at the iommu_domain level. __arm_smmu_sva_bind is therefore only called once for a device/mm pair.
--- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c | 12 +----------- 1 file changed, 1 insertion(+), 11 deletions(-)
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c +1 -11
@@ -29,7 +29,6 @@ struct arm_smmu_bond {struct mm_struct *mm;struct arm_smmu_mmu_notifier *smmu_mn;struct list_head list; -refcount_t refs; };#define sva_to_bond (handle) \
@@ -377,21 +376,12 @@ __arm_smmu_sva_bind(struct device *dev, struct mm_struct *mm)if (!master || !master->sva_enabled)return ERR_PTR (-ENODEV ); -/* If bind() was already called for this {dev, mm} pair, reuse it. */ -list_for_each_entry (bond, &master->bonds, list) { -if (bond->mm == mm) { -refcount_inc (&bond->refs); -return &bond->sva; - } - } - bond =kzalloc (sizeof (*bond),GFP_KERNEL );if (!bond)return ERR_PTR (-ENOMEM ); bond->mm = mm; bond->sva.dev = dev; -refcount_set (&bond->refs,1 ); bond->smmu_mn =arm_smmu_mmu_notifier_get (smmu_domain, mm);if (IS_ERR (bond->smmu_mn)) {
@@ -570,7 +560,7 @@ void arm_smmu_sva_remove_dev_pasid(struct iommu_domain *domain, } } -if (!WARN_ON (!bond) &&refcount_dec_and_test (&bond->refs)) { +if (!WARN_ON (!bond)) {list_del (&bond->list);arm_smmu_mmu_notifier_put (bond->smmu_mn);kfree (bond);
Signature · 1 line
2.41.0.162.gfafddb0af9-goog
PATCH v4 11/13 iommu/arm-smmu-v3-sva: Clean unused iommu_sva
In reply to Michael ShavitThe __arm_smmu_sva_bind function returned an unused iommu_sva handle that can be removed.
--- .../iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c | 18 ++++++------------ 1 file changed, 6 insertions(+), 12 deletions(-)
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c +6 -12
@@ -25,7 +25,6 @@ struct arm_smmu_mmu_notifier {#define mn_to_smmu (mn) container_of(mn, struct arm_smmu_mmu_notifier, mn)struct arm_smmu_bond { -struct iommu_sva sva;struct mm_struct *mm;struct arm_smmu_mmu_notifier *smmu_mn;struct list_head list;
@@ -364,8 +363,7 @@ static void arm_smmu_mmu_notifier_put(struct arm_smmu_mmu_notifier *smmu_mn)arm_smmu_free_shared_cd (cd); } -static struct iommu_sva * -__arm_smmu_sva_bind (struct device *dev,struct mm_struct *mm) +static int __arm_smmu_sva_bind (struct device *dev,struct mm_struct *mm) {int ret;struct arm_smmu_bond *bond;
@@ -374,14 +372,13 @@ __arm_smmu_sva_bind(struct device *dev, struct mm_struct *mm)struct arm_smmu_domain *smmu_domain =to_smmu_domain (domain);if (!master || !master->sva_enabled) -return ERR_PTR (-ENODEV ); +return -ENODEV ; bond =kzalloc (sizeof (*bond),GFP_KERNEL );if (!bond) -return ERR_PTR (-ENOMEM ); +return -ENOMEM ; bond->mm = mm; - bond->sva.dev = dev; bond->smmu_mn =arm_smmu_mmu_notifier_get (smmu_domain, mm);if (IS_ERR (bond->smmu_mn)) {
@@ -390,11 +387,11 @@ __arm_smmu_sva_bind(struct device *dev, struct mm_struct *mm) }list_add (&bond->list, &master->bonds); -return &bond->sva; +return 0 ; err_free_bond:kfree (bond); -return ERR_PTR (ret); +return ret; }bool arm_smmu_sva_supported (struct arm_smmu_device *smmu)
@@ -572,13 +569,10 @@ static int arm_smmu_sva_set_dev_pasid(struct iommu_domain *domain, struct device *dev, ioasid_t id) { int ret = 0; -struct iommu_sva *handle;struct mm_struct *mm = domain->mm;mutex_lock (&sva_lock); - handle =__arm_smmu_sva_bind (dev, mm); -if (IS_ERR (handle)) - ret =PTR_ERR (handle); + ret =__arm_smmu_sva_bind (dev, mm);mutex_unlock (&sva_lock);return ret;
Signature · 1 line
2.41.0.162.gfafddb0af9-goog
PATCH v4 12/13 iommu/arm-smmu-v3-sva: Remove arm_smmu_bond
In reply to Michael ShavitThere's a 1:1 relationship between arm_smmu_bond and the iommu_domain used in set_dev_pasid/remove_dev_pasid. arm_smmu_bond has become an unnecessary complication. It's more natural to store any needed information at the iommu_domain container level.
--- .../iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c | 69 +++++++------------ drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 1 - drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h | 2 +- 3 files changed, 24 insertions(+), 48 deletions(-)
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c +23 -46
@@ -24,14 +24,13 @@ struct arm_smmu_mmu_notifier {#define mn_to_smmu (mn) container_of(mn, struct arm_smmu_mmu_notifier, mn) -struct arm_smmu_bond { -struct mm_struct *mm; +struct arm_smmu_sva_domain { +struct iommu_domain iommu_domain;struct arm_smmu_mmu_notifier *smmu_mn; -struct list_head list; }; -#define sva_to_bond (handle) \ - container_of(handle, struct arm_smmu_bond, sva) +#define to_sva_domain (domain) \ + container_of(domain, struct arm_smmu_sva_domain, iommu_domain)static DEFINE_MUTEX (sva_lock);
@@ -363,10 +362,10 @@ static void arm_smmu_mmu_notifier_put(struct arm_smmu_mmu_notifier *smmu_mn)arm_smmu_free_shared_cd (cd); } -static int __arm_smmu_sva_bind (struct device *dev,struct mm_struct *mm) +static int __arm_smmu_sva_bind (struct device *dev, +struct arm_smmu_sva_domain *sva_domain, +struct mm_struct *mm) { -int ret; -struct arm_smmu_bond *bond;struct arm_smmu_master *master =dev_iommu_priv_get (dev);struct iommu_domain *domain =iommu_get_domain_for_dev (dev);struct arm_smmu_domain *smmu_domain =to_smmu_domain (domain);
@@ -374,24 +373,14 @@ static int __arm_smmu_sva_bind(struct device *dev, struct mm_struct *mm)if (!master || !master->sva_enabled)return -ENODEV ; - bond =kzalloc (sizeof (*bond),GFP_KERNEL ); -if (!bond) -return -ENOMEM ; - - bond->mm = mm; - - bond->smmu_mn =arm_smmu_mmu_notifier_get (smmu_domain, mm); -if (IS_ERR (bond->smmu_mn)) { - ret =PTR_ERR (bond->smmu_mn); - goto err_free_bond; + sva_domain->smmu_mn =arm_smmu_mmu_notifier_get (smmu_domain, + mm); +if (IS_ERR (sva_domain->smmu_mn)) { + sva_domain->smmu_mn =NULL ; +return PTR_ERR (sva_domain->smmu_mn); } - -list_add (&bond->list, &master->bonds); + master->nr_attached_sva_domains +=1 ;return 0 ; - -err_free_bond: -kfree (bond); -return ret; }bool arm_smmu_sva_supported (struct arm_smmu_device *smmu)
@@ -521,7 +510,7 @@ int arm_smmu_master_enable_sva(struct arm_smmu_master *master)int arm_smmu_master_disable_sva (struct arm_smmu_master *master) {mutex_lock (&sva_lock); -if (!list_empty (&master->bonds)) { +if (master->nr_attached_sva_domains !=0 ) {dev_err (master->dev,"cannot disable SVA, device is bound\n" );mutex_unlock (&sva_lock);return -EBUSY ;
@@ -545,23 +534,12 @@ void arm_smmu_sva_notifier_synchronize(void)void arm_smmu_sva_remove_dev_pasid (struct iommu_domain *domain,struct device *dev,ioasid_t id) { -struct mm_struct *mm = domain->mm; -struct arm_smmu_bond *bond =NULL , *t; +struct arm_smmu_sva_domain *sva_domain =to_sva_domain (domain);struct arm_smmu_master *master =dev_iommu_priv_get (dev);mutex_lock (&sva_lock); -list_for_each_entry (t, &master->bonds, list) { -if (t->mm == mm) { - bond = t; -break ; - } - } - -if (!WARN_ON (!bond)) { -list_del (&bond->list); -arm_smmu_mmu_notifier_put (bond->smmu_mn); -kfree (bond); - } + master->nr_attached_sva_domains -=1 ; +arm_smmu_mmu_notifier_put (sva_domain->smmu_mn);mutex_unlock (&sva_lock); }
@@ -572,7 +550,7 @@ static int arm_smmu_sva_set_dev_pasid(struct iommu_domain *domain,struct mm_struct *mm = domain->mm;mutex_lock (&sva_lock); - ret =__arm_smmu_sva_bind (dev, mm); + ret =__arm_smmu_sva_bind (dev,to_sva_domain (domain), mm);mutex_unlock (&sva_lock);return ret;
@@ -590,12 +568,11 @@ static const struct iommu_domain_ops arm_smmu_sva_domain_ops = {struct iommu_domain *arm_smmu_sva_domain_alloc (void ) { -struct iommu_domain *domain; +struct arm_smmu_sva_domain *sva_domain; - domain =kzalloc (sizeof (*domain),GFP_KERNEL ); -if (!domain) + sva_domain =kzalloc (sizeof (*sva_domain),GFP_KERNEL ); +if (!sva_domain)return NULL ; - domain->ops = &arm_smmu_sva_domain_ops; - -return domain; + sva_domain->iommu_domain.ops = &arm_smmu_sva_domain_ops; +return &sva_domain->iommu_domain; }
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c +0 -1
@@ -2784,7 +2784,6 @@ static struct iommu_device *arm_smmu_probe_device(struct device *dev) master->dev = dev; master->smmu = smmu; -INIT_LIST_HEAD (&master->bonds);dev_iommu_priv_set (dev, master); ret =arm_smmu_insert_master (smmu, master);
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h +1 -1
@@ -705,7 +705,7 @@ struct arm_smmu_master {bool stall_enabled;bool sva_enabled;bool iopf_enabled; -struct list_head bonds; +unsigned int nr_attached_sva_domains;unsigned int ssid_bits;unsigned int nr_attached_pasid_domains; };
Signature · 1 line
2.41.0.162.gfafddb0af9-goog
Re: [PATCH v4 12/13] iommu/arm-smmu-v3-sva: Remove arm_smmu_bond
Quoted 17 lines · Michael Shavit
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c +1
-8@@ -545,23 +534,12 @@ void arm_smmu_sva_notifier_synchronize(void)void arm_smmu_sva_remove_dev_pasid (struct iommu_domain *domain,struct device *dev,ioasid_t id) { -struct mm_struct *mm = domain->mm; -struct arm_smmu_bond *bond =NULL , *t; +struct arm_smmu_sva_domain *sva_domain =to_sva_domain (domain);struct arm_smmu_master *master =dev_iommu_priv_get (dev);mutex_lock (&sva_lock); -list_for_each_entry (t, &master->bonds, list) { -if (t->mm == mm) { - bond = t; -break ; - } - }
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c void arm_smmu_sva_notifier_synchronize(void) · line 544
Quoted 10 lines
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c +2
-6- -if (!WARN_ON (!bond)) { -list_del (&bond->list); -arm_smmu_mmu_notifier_put (bond->smmu_mn); -kfree (bond); - } + master->nr_attached_sva_domains -=1 ; +arm_smmu_mmu_notifier_put (sva_domain->smmu_mn);mutex_unlock (&sva_lock); }
I've encountered a bug with this change while working on a follow-up series. arm_smmu_sva_remove_dev_pasid() is called on the domain if arm_smmu_sva_set_dev_pasid failed. Removing the WARN_ON(!bond) condition removed a protection against that scenario. I have a fix for this change that I'll push into a v5 of the series.
PATCH v4 13/13 iommu/arm-smmu-v3-sva: Add check when enabling sva
In reply to Michael ShavitSVA domains can only be attached when the master's STEs have a stage 1 domain.
--- drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c | 6 ++++++ 1 file changed, 6 insertions(+)
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3-sva.c +6 -0
@@ -499,9 +499,15 @@ int arm_smmu_master_enable_sva(struct arm_smmu_master *master)int ret;mutex_lock (&sva_lock); + +if (!master->s1_cfg) { + ret = -EBUSY ; + goto unlock; + } ret =arm_smmu_master_sva_enable_iopf (master);if (!ret) master->sva_enabled = true; +unlock:mutex_unlock (&sva_lock);return ret;
Signature · 1 line
2.41.0.162.gfafddb0af9-goog