iommu/amd: Enable page-selective flushes
Subsystems (4)
cover iommu/amd: Enable page-selective flushes
From: Nadav Amit <[email protected]>
The previous patch, commit 268aa4548277 ("iommu/amd: Page-specific invalidations for more than one page") was supposed to enable page-selective IOTLB flushes on AMD.
The patch had an embaressing bug, and I apologize for it.
Analysis as for why this bug did not result in failures raised additional issues that caused at least most of the IOTLB flushes not to be page-selective ones.
The first patch corrects the bug from the previous patch. The next patches enable page-selective invalidations, which were not enabled despite the previous patch.
Nadav Amit (4):
iommu/amd: Fix wrong parentheses on page-specific invalidations iommu/amd: Selective flush on unmap
iommu/amd: Do not sync on page size changes iommu/amd: Do not use flush-queue when NpCache is on
drivers/iommu/amd/init.c | 7 ++++++- drivers/iommu/amd/iommu.c | 18 +++++++++++++++--- include/linux/iommu.h | 3 ++- 3 files changed, 23 insertions(+), 5 deletions(-)
Signature · 6 lines
2.25.1 _______________________________________________ iommu mailing list [email protected] https://lists.linuxfoundation.org/mailman/listinfo/iommu
PATCH 1/4 iommu/amd: Fix wrong parentheses on page-specific invalidations
From: Nadav Amit <[email protected]>
The logic to determine the mask of page-specific invalidations was tested in userspace. As the code was copied into the kernel, the parentheses were mistakenly set in the wrong place, resulting in the wrong mask.
Fix it.
--- drivers/iommu/amd/iommu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/iommu/amd/iommu.c +1 -1
@@ -884,7 +884,7 @@ static inline u64 build_inv_address(u64 address, size_t size)* The msb-bit must be clear on the address. Just set all the * lower bits. */ - address |=1ull << (msb_diff -1 ); + address |= (1ull << msb_diff) -1 ; }/* Clear bits 11:0 */
Signature · 6 lines
2.25.1 _______________________________________________ iommu mailing list [email protected] https://lists.linuxfoundation.org/mailman/listinfo/iommu
Re: [PATCH 1/4] iommu/amd: Fix wrong parentheses on page-specific invalidations
Quoted 16 lines · Nadav Amit
From: Nadav Amit <[email protected]>
The logic to determine the mask of page-specific invalidations was tested in userspace. As the code was copied into the kernel, the parentheses were mistakenly set in the wrong place, resulting in the wrong mask.
Fix it.
Cc Joerg Roedel <[email protected]>Cc Will Deacon <[email protected]>Cc Jiajun Cao <[email protected]>Fixes 268aa4548277 ("iommu/amd: Page-specific invalidations for more than one page")Signed-off-by Nadav Amit <[email protected]>
Quoted 3 lines
--- drivers/iommu/amd/iommu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
Applied this one for v5.13, thanks Nadav.
Somehow the rest of the patch-set got screwed up during sending or so, at least I see some patches twice in my inbox and with different subjects.
Can you please re-send patches 2-4 when -rc3 it out?
Thanks,
Joerg
Re: [PATCH 1/4] iommu/amd: Fix wrong parentheses on page-specific invalidations
Quoted 25 lines · Joerg Roedel
Quoted 16 lines · Nadav Amit
From: Nadav Amit <[email protected]>
The logic to determine the mask of page-specific invalidations was tested in userspace. As the code was copied into the kernel, the parentheses were mistakenly set in the wrong place, resulting in the wrong mask.
Fix it.
Cc Joerg Roedel <[email protected]>Cc Will Deacon <[email protected]>Cc Jiajun Cao <[email protected]>Fixes 268aa4548277 ("iommu/amd: Page-specific invalidations for more than one page")Signed-off-by Nadav Amit <[email protected]>Quoted 3 lines
---
drivers/iommu/amd/iommu.c| 2 +-1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 7 lines
Applied this one for v5.13, thanks Nadav.
Somehow the rest of the patch-set got screwed up during sending or so, at least I see some patches twice in my inbox and with different subjects.
Can you please re-send patches 2-4 when -rc3 it out?
Joerg,
Thanks for your understanding. I sent a version based on -rc3 a week ago.
I noticed that there was some confusion regarding rc numbers. Do you need a new version based on rc4 or can you apply the version I sent?
Thanks,
Nadav
_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
PATCH 2/4 iommu/amd: Do not sync on page size changes
In reply to Nadav AmitFrom: Nadav Amit <[email protected]>
Some IOMMU architectures perform invalidations regardless of the page size. In such architectures there is no need to sync when the page size changes. In such architecture, there is no need to regard pgsize when making interim flush in iommu_iotlb_gather_add_page().
Add a "no_sync_on_pgsize_change" property for each IOMMU ops to decide whether gather's pgsize is tracked and triggers a flush.
--- drivers/iommu/amd/iommu.c | 1 + include/linux/iommu.h | 3 ++- 2 files changed, 3 insertions(+), 1 deletion(-)
drivers/iommu/amd/iommu.c +1 -0
@@ -2204,6 +2204,7 @@ const struct iommu_ops amd_iommu_ops = { .put_resv_regions = generic_iommu_put_resv_regions, .is_attach_deferred = amd_iommu_is_attach_deferred, .pgsize_bitmap = AMD_IOMMU_PGSIZES, + .ignore_gather_pgsize = true, .flush_iotlb_all = amd_iommu_flush_iotlb_all, .iotlb_sync = amd_iommu_iotlb_sync, .def_domain_type = amd_iommu_def_domain_type,
include/linux/iommu.h +2 -1
@@ -284,6 +284,7 @@ struct iommu_ops {int (*def_domain_type)(struct device *dev);unsigned long pgsize_bitmap; +bool ignore_gather_pgsize;struct module *owner; };
@@ -508,7 +509,7 @@ static inline void iommu_iotlb_gather_add_page(struct iommu_domain *domain,* a different granularity, then sync the TLB so that the gather * structure can be rewritten. */ -if (gather->pgsize != size || +if ((gather->pgsize != size && !domain->ops->ignore_gather_pgsize) || end +1 < gather->start || start > gather->end +1 ) {if (gather->pgsize)iommu_iotlb_sync (domain, gather);
Signature · 6 lines
2.25.1 _______________________________________________ iommu mailing list [email protected] https://lists.linuxfoundation.org/mailman/listinfo/iommu
PATCH 2/4 iommu/amd: Selective flush on unmap
In reply to Nadav AmitFrom: Nadav Amit <[email protected]>
Recent patch attempted to enable selective page flushes on AMD IOMMU but neglected to adapt amd_iommu_iotlb_sync() to use the selective flushes.
Adapt amd_iommu_iotlb_sync() to use selective flushes and change amd_iommu_unmap() to collect the flushes. As a defensive measure, to avoid potential issues as those that the Intel IOMMU driver encountered recently, flush the page-walk caches by always setting the "pde" parameter. This can be removed later.
--- drivers/iommu/amd/iommu.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-)
drivers/iommu/amd/iommu.c +13 -2
@@ -2057,12 +2057,17 @@ static size_t amd_iommu_unmap(struct iommu_domain *dom, unsigned long iova, {struct protection_domain *domain =to_pdomain (dom);struct io_pgtable_ops *ops = &domain->iop.iop.ops; +size_t r;if ((amd_iommu_pgtable ==AMD_IOMMU_V1 ) && (domain->iop.mode ==PAGE_MODE_NONE ))return 0 ; -return (ops->unmap) ? ops->unmap (ops, iova, page_size, gather) :0 ; + r = (ops->unmap) ? ops->unmap (ops, iova, page_size, gather) :0 ; + +iommu_iotlb_gather_add_page (dom, gather, iova, page_size); + +return r; } static phys_addr_t amd_iommu_iova_to_phys(struct iommu_domain *dom,
@@ -2165,7 +2170,13 @@ static void amd_iommu_flush_iotlb_all(struct iommu_domain *domain)static void amd_iommu_iotlb_sync (struct iommu_domain *domain,struct iommu_iotlb_gather *gather) { -amd_iommu_flush_iotlb_all (domain); +struct protection_domain *dom =to_pdomain (domain); +unsigned long flags; + +spin_lock_irqsave (&dom->lock, flags); +__domain_flush_pages (dom, gather->start, gather->end - gather->start,1 ); +amd_iommu_domain_flush_complete (dom); +spin_unlock_irqrestore (&dom->lock, flags); }static int amd_iommu_def_domain_type (struct device *dev)
Signature · 6 lines
2.25.1 _______________________________________________ iommu mailing list [email protected] https://lists.linuxfoundation.org/mailman/listinfo/iommu
PATCH 3/4 iommu/amd: Do not sync on page size changes
In reply to Nadav AmitFrom: Nadav Amit <[email protected]>
Some IOMMU architectures perform invalidations regardless of the page size. In such architectures there is no need to sync when the page size changes or to regard pgsize when making interim flush in iommu_iotlb_gather_add_page().
Add a "ignore_gather_pgsize" property for each IOMMU-ops to decide whether gather's pgsize is tracked and triggers a flush.
--- drivers/iommu/amd/iommu.c | 1 + include/linux/iommu.h | 3 ++- 2 files changed, 3 insertions(+), 1 deletion(-)
drivers/iommu/amd/iommu.c +1 -0
@@ -2215,6 +2215,7 @@ const struct iommu_ops amd_iommu_ops = { .put_resv_regions = generic_iommu_put_resv_regions, .is_attach_deferred = amd_iommu_is_attach_deferred, .pgsize_bitmap = AMD_IOMMU_PGSIZES, + .ignore_gather_pgsize = true, .flush_iotlb_all = amd_iommu_flush_iotlb_all, .iotlb_sync = amd_iommu_iotlb_sync, .def_domain_type = amd_iommu_def_domain_type,
include/linux/iommu.h +2 -1
@@ -284,6 +284,7 @@ struct iommu_ops {int (*def_domain_type)(struct device *dev);unsigned long pgsize_bitmap; +bool ignore_gather_pgsize;struct module *owner; };
@@ -508,7 +509,7 @@ static inline void iommu_iotlb_gather_add_page(struct iommu_domain *domain,* a different granularity, then sync the TLB so that the gather * structure can be rewritten. */ -if (gather->pgsize != size || +if ((gather->pgsize != size && !domain->ops->ignore_gather_pgsize) || end +1 < gather->start || start > gather->end +1 ) {if (gather->pgsize)iommu_iotlb_sync (domain, gather);
Signature · 6 lines
2.25.1 _______________________________________________ iommu mailing list [email protected] https://lists.linuxfoundation.org/mailman/listinfo/iommu
Robin Murphy
reviewerRe: [PATCH 3/4] iommu/amd: Do not sync on page size changes
Quoted 9 lines · Nadav Amit
From: Nadav Amit <[email protected]>
Some IOMMU architectures perform invalidations regardless of the page size. In such architectures there is no need to sync when the page size changes or to regard pgsize when making interim flush in
iommu_iotlb_gather_add_page().Add a "
ignore_gather_pgsize" property for each IOMMU-ops to decide whether gather's pgsize is tracked and triggers a flush.
I've objected before[1], and I'll readily object again ;)
I still think it's very silly to add a bunch of indirection all over the place to make a helper function not do the main thing it's intended to help with. If you only need trivial address gathering then it's far simpler to just implement trivial address gathering. I suppose if you really want to you could factor out another helper to share the 5 lines of code which ended up in mtk-iommu (see commit f21ae3b10084).
Robin.
[1] https://lore.kernel.org/linux-iommu/[email protected]/
Quoted 45 lines
Cc Joerg Roedel <[email protected]>Cc Will Deacon <[email protected]>Cc Jiajun Cao <[email protected]>Signed-off-by Nadav Amit <[email protected]>--- drivers/iommu/amd/iommu.c | 1 + include/linux/iommu.h | 3 ++- 2 files changed, 3 insertions(+), 1 deletion(-)drivers/iommu/amd/iommu.c +1
-0@@ -2215,6 +2215,7 @@ const struct iommu_ops amd_iommu_ops = { .put_resv_regions = generic_iommu_put_resv_regions, .is_attach_deferred = amd_iommu_is_attach_deferred, .pgsize_bitmap = AMD_IOMMU_PGSIZES, + .ignore_gather_pgsize = true, .flush_iotlb_all = amd_iommu_flush_iotlb_all, .iotlb_sync = amd_iommu_iotlb_sync, .def_domain_type = amd_iommu_def_domain_type,include/linux/iommu.h +2
-1@@ -284,6 +284,7 @@ struct iommu_ops {int (*def_domain_type)(struct device *dev);unsigned long pgsize_bitmap; +bool ignore_gather_pgsize;struct module *owner; };@@ -508,7 +509,7 @@ static inline void iommu_iotlb_gather_add_page(struct iommu_domain *domain,* a different granularity, then sync the TLB so that the gather * structure can be rewritten. */ -if (gather->pgsize != size || +if ((gather->pgsize != size && !domain->ops->ignore_gather_pgsize) || end +1 < gather->start || start > gather->end +1 ) {if (gather->pgsize)iommu_iotlb_sync (domain, gather);
Re: [PATCH 3/4] iommu/amd: Do not sync on page size changes
Quoted 13 lines · Robin Murphy
Quoted 7 lines · Nadav Amit
From: Nadav Amit <[email protected]>
Some IOMMU architectures perform invalidations regardless of the page size. In such architectures there is no need to sync when the page size changes or to regard pgsize when making interim flush iniommu_iotlb_gather_add_page().
Add a "ignore_gather_pgsize" property for each IOMMU-ops to decide whether gather's pgsize is tracked and triggers a flush.
Quoted 3 lines
I've objected before[1], and I'll readily object again ;)
I still think it's very silly to add a bunch of indirection all over the place to make a helper function not do the main thing it's intended to help with. If you only need trivial address gathering then it's far simpler to just implement trivial address gathering. I suppose if you really want to you could factor out another helper to share the 5 lines of code which ended up in mtk-iommu (see commit
f21ae3b10084).
Thanks, Robin.
I read your comments but I cannot fully understand the alternative that you propose, although I do understand your objection to the indirection “ignore_gather_pgsize”. Would it be ok if “ignore_gather_pgsize" was provided as an argument for iommu_iotlb_gather_add_page()?
In general, I can live without this patch. It probably would have negligent impact on performance anyhow.
Regards,
Nadav
Robin Murphy
reviewerRe: [PATCH 3/4] iommu/amd: Do not sync on page size changes
Quoted 13 lines · Nadav Amit
Quoted 13 lines · Robin Murphy
Quoted 7 lines · Nadav Amit
From: Nadav Amit <[email protected]>
Some IOMMU architectures perform invalidations regardless of the page size. In such architectures there is no need to sync when the page size changes or to regard pgsize when making interim flush iniommu_iotlb_gather_add_page().
Add a "ignore_gather_pgsize" property for each IOMMU-ops to decide whether gather's pgsize is tracked and triggers a flush.
Quoted 7 lines
Quoted 3 lines
I've objected before[1], and I'll readily object again ;)
I still think it's very silly to add a bunch of indirection all over the place to make a helper function not do the main thing it's intended to help with. If you only need trivial address gathering then it's far simpler to just implement trivial address gathering. I suppose if you really want to you could factor out another helper to share the 5 lines of code which ended up in mtk-iommu (see commit
f21ae3b10084).Thanks, Robin.
I read your comments but I cannot fully understand the alternative that you propose, although I do understand your objection to the indirection “ignore_gather_pgsize”. Would it be ok if “ignore_gather_pgsize" was provided as an argument for
iommu_iotlb_gather_add_page()?
No, I mean if iommu_iotlb_gather_add_page() doesn't have the behaviour your driver wants, just don't call it. Write or factor out a suitable helper that *does* do what you want and call that, or just implement the logic directly inline. Indirect argument or not, it just doesn't make much sense to have a helper function call which says "do this except don't do most of it".
Quoted 1 line
In general, I can live without this patch. It probably would have negligent impact on performance anyhow.
As I implied, it sounds like your needs are the same as the Mediatek driver had, so you could readily factor out a new page-size-agnostic gather helper from that. I fully support making the functional change to amd-iommu *somehow* - nobody likes unnecessary syncs - just not with this particular implementation :)
Robin.
Re: [PATCH 3/4] iommu/amd: Do not sync on page size changes
Quoted 19 lines · Robin Murphy
Quoted 14 lines · Nadav Amit
Quoted 14 lines · Robin Murphy
Quoted 7 lines · Nadav Amit
From: Nadav Amit <[email protected]>
Some IOMMU architectures perform invalidations regardless of the page size. In such architectures there is no need to sync when the page size changes or to regard pgsize when making interim flush iniommu_iotlb_gather_add_page().
Add a "ignore_gather_pgsize" property for each IOMMU-ops to decide whether gather's pgsize is tracked and triggers a flush.I've objected before[1], and I'll readily object again ;)
I still think it's very silly to add a bunch of indirection all over the place to make a helper function not do the main thing it's intended to help with. If you only need trivial address gathering then it's far simpler to just implement trivial address gathering. I suppose if you really want to you could factor out another helper to share the 5 lines of code which ended up in mtk-iommu (see commit
f21ae3b10084).
Quoted 8 lines
Quoted 2 lines
Thanks, Robin.
I read your comments but I cannot fully understand the alternative that you propose, although I do understand your objection to the indirection “ignore_gather_pgsize”. Would it be ok if “ignore_gather_pgsize" was provided as an argument foriommu_iotlb_gather_add_page()?No, I mean if
iommu_iotlb_gather_add_page()doesn't have the behaviour your driver wants, just don't call it. Write or factor out a suitable helper that *does* do what you want and call that, or just implement the logic directly inline. Indirect argument or not, it just doesn't make much sense to have a helper function call which says "do this except don't do most of it".Quoted 1 line
In general, I can live without this patch. It probably would have negligent impact on performance anyhow.
As I implied, it sounds like your needs are the same as the Mediatek driver had, so you could readily factor out a new page-size-agnostic gather helper from that. I fully support making the functional change to amd-iommu *somehow* - nobody likes unnecessary syncs - just not with this particular implementation :)
Hm… avoid code duplication I need to extract some common code to another function.
Is the following resembles what you had in mind (untested)?
-- >8 --
Subject: [PATCH] iommu: add iommu_iotlb_gather_add_page_ignore_pgsize()
--- drivers/iommu/mtk_iommu.c | 7 ++--- include/linux/iommu.h | 55 ++++++++++++++++++++++++++++++--------- 2 files changed, 44 insertions(+), 18 deletions(-)
drivers/iommu/mtk_iommu.c +2 -5
@@ -520,12 +520,9 @@ static size_t mtk_iommu_unmap(struct iommu_domain *domain, struct iommu_iotlb_gather *gather) {struct mtk_iommu_domain *dom =to_mtk_domain (domain); -unsigned long end = iova + size -1 ; -if (gather->start > iova) - gather->start = iova; -if (gather->end < end) - gather->end = end; +iommu_iotlb_gather_update_range (gather, iova, size); +return dom->iop->unmap (dom->iop, iova, size, gather); }
include/linux/iommu.h +42 -13
71 lines not on this page · raw message
--
2.25.1
PATCH 3/4 iommu/amd: Selective flush on unmap
In reply to Nadav AmitFrom: Nadav Amit <[email protected]>
Recent patch attempted to enable selective page flushes on AMD IOMMU but neglected to adapt amd_iommu_iotlb_sync() to use the selective flushes.
Adapt amd_iommu_iotlb_sync() to use selective flushes and change amd_iommu_unmap() to collect the flushes. As a defensive measure, to avoid potential issues as those that the Intel IOMMU driver encountered recently, flush the page-walk caches by always setting the "pde" parameter. This can be removed later.
--- drivers/iommu/amd/iommu.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-)
drivers/iommu/amd/iommu.c +13 -2
@@ -2057,12 +2057,17 @@ static size_t amd_iommu_unmap(struct iommu_domain *dom, unsigned long iova, {struct protection_domain *domain =to_pdomain (dom);struct io_pgtable_ops *ops = &domain->iop.iop.ops; +size_t r;if ((amd_iommu_pgtable ==AMD_IOMMU_V1 ) && (domain->iop.mode ==PAGE_MODE_NONE ))return 0 ; -return (ops->unmap) ? ops->unmap (ops, iova, page_size, gather) :0 ; + r = (ops->unmap) ? ops->unmap (ops, iova, page_size, gather) :0 ; + +iommu_iotlb_gather_add_page (dom, gather, iova, page_size); + +return r; } static phys_addr_t amd_iommu_iova_to_phys(struct iommu_domain *dom,
@@ -2165,7 +2170,13 @@ static void amd_iommu_flush_iotlb_all(struct iommu_domain *domain)static void amd_iommu_iotlb_sync (struct iommu_domain *domain,struct iommu_iotlb_gather *gather) { -amd_iommu_flush_iotlb_all (domain); +struct protection_domain *dom =to_pdomain (domain); +unsigned long flags; + +spin_lock_irqsave (&dom->lock, flags); +__domain_flush_pages (dom, gather->start, gather->end - gather->start,1 ); +amd_iommu_domain_flush_complete (dom); +spin_unlock_irqrestore (&dom->lock, flags); }static int amd_iommu_def_domain_type (struct device *dev)
Signature · 6 lines
2.25.1 _______________________________________________ iommu mailing list [email protected] https://lists.linuxfoundation.org/mailman/listinfo/iommu
PATCH 4/4 iommu/amd: Do not use flush-queue when NpCache is on
In reply to Nadav AmitFrom: Nadav Amit <[email protected]>
Do not use flush-queue on virtualized environments, where the NpCache capability of the IOMMU is set. This is required to reduce virtualization overheads.
This change follows a similar change to Intel's VT-d and a detailed explanation as for the rationale is described in commit 29b32839725f ("iommu/vt-d: Do not use flush-queue when caching-mode is on").
--- drivers/iommu/amd/init.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)
drivers/iommu/amd/init.c +6 -1
@@ -1850,8 +1850,13 @@ static int __init iommu_init_pci(struct amd_iommu *iommu)if (ret)return ret; -if (iommu->cap & (1UL <<IOMMU_CAP_NPCACHE )) +if (iommu->cap & (1UL <<IOMMU_CAP_NPCACHE )) { +if (!amd_iommu_unmap_flush) +pr_warn_once ("IOMMU batching is disabled due to virtualization" ); + amd_iommu_np_cache = true; + amd_iommu_unmap_flush = true; + }init_iommu_perf_ctr (iommu);
Signature · 6 lines
2.25.1 _______________________________________________ iommu mailing list [email protected] https://lists.linuxfoundation.org/mailman/listinfo/iommu