♟ br

Code deduplication: use SID-getting functions in smmu_find().

Split-out SID-getting code to a separate functions.

Allocate resources selectively.

Change __unused to __diagused.

Add FDT attachment driver for ARM System MMU v3.2 controller.

Change iommu list lock to SX as we need a sleepable lock so we

Fix !INVARIANTS build.

Check if iommu ctx is valid before passing it to functions.

Return error code from IOMMU_MAP and fix build.

Add OFW_IOMMU id type for pci_get_id() so we can ask parent IOMMU

rename PCI_ID_IOMMU to PCI_ID_OWF_IOMMU

remove committed part

Fix ofw_bus_iommu_map() since there is no limit on amount of maps

Add ofw_bus_iommu_map() that maps PCI requester ID to an IOMMU

Reuse pci_get_id() method, just add IOMMU id type.

Remove cdns_i2c from files

In D34262#775530, @andrew wrote:

Where did this come from? (i.e. which git tree).

Are we expecting to continue development in FreeBSD, or should it be considered contrib code & be a vendor import?

arm/pmu: add ACPI attachment.

Collect all PPIs since they indeed could be different on different CPUs

Fix bug: Initialize cpuid to -1.
Also for PPI we need just one resource, so stop iteration over ACPI tables if PMU PPI interrupt found.

In D24423#737819, @andrew wrote:

Was it intentional you reverted MAX_RLEN back to 8?

Restore MAX_RLEN change

Use intr_is_per_cpu()

Address jrtc27's and mhorne's notes

I've tested this on ARM Neoverse N1 (ARM N1SDP) and it works fine (both sampling and counting modes).

o Since INTRNG abuses 'virtual' property of a resource to store interrupt config, use it to set desired interrupt configuration here.
o Bump MAX_RLEN to MAXCPU

Prevent repeated deallocation of a resource.

I like previous version, because it is easier to read the diff

In D32325#729404, @jrtc27 wrote:

I'm not sure this makes sense to upstream? And it'll conflict with whatever actual IOMMU spec ends up being added to RISC-V when they finally get round to it.

hdspe: Support for mono sound.

o Fix i2c read operation for large transfers (more than 32 bytes).

fix transfer_len for reading as well. I think eliminate a need of delay

o Consider amount of bytes already transferred
o Add a small delay (not sure why it is needed, maybe another bug)

ofw: support for a single 'port' DTS property.

Move IOMMU code to a separate pmap module and switch ARM System MMU

Consider the broken card detect flag that comes from 'broken-cd;'

Use broken-cd property

remove unused headers

fix copyrights as code was copy-pasted partially from work by andy

add pinit/release prototypes

so this won't work when CPU page size and IOMMU page sizes are different

In D29731#667264, @markj wrote:

Can you provide any pointers to docs or driver code that illustrates how the functions are used?

Rename pmap_genter/gremove to pmap_gpu_enter/remove

cast vm_offset_t to dcache_wb

no need to invalidate ptes just write-back

I also added SMP support to Unleashed board. But that may not be notable change :)

br (Ruslan Bukin)
User

Projects

User Details

Recent Activity
View All

Tue, May 10

Mon, May 9

Sat, May 7

Fri, May 6

Thu, May 5

Mar 10 2022

Feb 25 2022

Feb 15 2022

Feb 14 2022

Feb 13 2022

Nov 5 2021

Nov 2 2021

Oct 27 2021

Oct 26 2021

Oct 12 2021

Oct 11 2021

Oct 5 2021

Jul 19 2021

May 27 2021

May 26 2021

May 25 2021

May 8 2021

May 7 2021

Apr 27 2021

Apr 26 2021

Apr 21 2021

Apr 16 2021

Apr 15 2021

Apr 14 2021

Apr 13 2021

Apr 12 2021

Apr 8 2021

br (Ruslan Bukin)User