[userspace LL] scheduling: LLEXT, DP & multicore by lyakh · Pull Request #10945 · thesofproject/sof

lyakh · 2026-06-18T16:08:25Z

This includes #10558 and my patches on top to enable LLEXT and multicore. Current status: passes simple tests with nocodec with both core 0 and core 1 streaming. 2 streams simultaneously run into a problem when the first of them terminates. WiP.

Update: now that #10558 has mostly been merged, this PR still carries remaining fixes for multicore, LLEXT, cold sections. At its peak this PR contained more than a 100 commits, as of today Aug 28th only 12 patches remain

Update: as of today Sep 2nd only 7 patches remain and this is now considered to be the final step (short of making userspace LL on by default on ACE3+)

lyakh · 2026-08-20T13:22:21Z

After multiple PRs got merged we're down from 108 commits to 32

lyakh · 2026-08-27T13:40:34Z

let me mark it "ready" to have some on-device testing

intel-sofci · 2026-09-01T13:05:40Z

PR 10945: test results

Run date: 2026-09-04 09:03 UTC

Tested commit: 6a70cdd8d32dfaab35b7dd602be7d6add461a79b

Copilot

🟡 Changes recommended

Confirmed issues include unsafe userspace LL domain mapping for DP modules and uninitialized/per-core IPC state leading to potential deadlocks or invalid memory access.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Review details

Suppressed comments (1)

src/ipc/ipc-common.c:619

ipc_user is allocated with sof_heap_alloc() but never zero-initialized; init_needed[] and the per-core thread/event pointers start as uninitialized heap garbage. This can cause ipc_user_forward_cmd() to deadlock (spurious init_needed[core]) or dereference invalid per-core pointers.

	struct ipc *ipc = ipc_get();
	struct ipc_user *ipc_user = sof_heap_alloc(sof_sys_user_heap_get(),
						   SOF_MEM_FLAG_USER | SOF_MEM_FLAG_COHERENT,
						   sizeof(*ipc_user), 0);
	int ret;

Files reviewed: 14/14 changed files
Comments generated: 2
Review effort level: Lite

kv2019i

Looks good, some minor comments and usual nagging about z_vrfy code ...

kv2019i · 2026-09-02T10:20:04Z

 #define IPC_TASK_POWERDOWN      BIT(3)

+#ifdef CONFIG_CORE_COUNT
+#define CORE_COUNT CONFIG_CORE_COUNT


Isn't CORE_COUNT a bit generic (and short) definition in a public header file?

Most places in SOF just use CONFIG_CORE_COUNT (defined in sof/src/platform/Kconfig), so this should be available for all builds. Maybe "ifndef CONFIG_CORE_COUNT" and set it 1 for the failing builds (library, cmocka, ztests?)

uhm, rather not. Really don't think #define CONFIG_foo ... is a good thing. How about IPC_DSP_CORE_COUNT?

@lyakh I'm good with IPC_DSP_ prefix. Or alternatively pull in some header that will ensure CONFIG_CORE_COUNT is defined.

kv2019i · 2026-09-02T10:35:13Z

    userspace/ksem.c
  )
-  if(CONFIG_USERSPACE AND CONFIG_SOF_USERSPACE_INTERFACE_ALLOC)
+  if(CONFIG_USERSPACE AND CONFIG_SOF_USERSPACE_INTERFACE_ALLOC AND NOT CONFIG_SOF_USERSPACE_LL)


Ack, filed #11147 to track this,

lyakh · 2026-09-03T12:11:29Z

@kv2019i comments addressed

lyakh · 2026-09-03T15:25:52Z

@lrudyX MTL test hung?

kv2019i · 2026-09-03T15:47:26Z

@lyakh wrote:

@lrudyX MTL test hung?

I restarted the MTL test.

Extract a privileged LLEXT-related part from lib_manager_module_create() into a separate function to be called from kernel context. At the same time lib_manager_mod_free_priv() already executes privileged operations; to make it callable in userspace, convert lib_manager_free_module() to a system call. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

If LL runs in userspace, it needs access to loaded LLEXT modules, running in DP more too. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

Make scheduling LL thread and synchronisation objects per-core and forward IPCs and scheduling events accordingly. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

Checking pointer for non-NULL after dereferencing it makes no sense. Swap the order. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

Simplify code by using cached core ID values instead of querying the hardware or re-evaluating expressions. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

In case of userspace LL scheduling the (also userspace) IPC thread needs access rights to DP assets like the thread itself and its stack and synchronisation primitives. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

When running in syscall context on behalf of a userspace thread dynamically mapped memory doesn't automatically become accessible. To make it accessible it has to be added to the thread memory domain. This is a problem for loadable modules with executable cold sections. To be able to execute them they have to be mapped to threads with the executable bit set. While for linking that memory has to be mapped writable. To solve the problem we perform linking from the kernel IPC context before forwarding to the userspace IPC thread. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

Re-enable LLEXT, DRAM execution and data, the DP scheduler and boot time testing with userspace LL with a small restriction for the latter - userspace allocation failure tests aren't currently working. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

lyakh force-pushed the llext-ull branch from 0c4e579 to 0f8ce76 Compare June 22, 2026 10:53

lyakh mentioned this pull request Jun 22, 2026

userspace: preparatory patches for multicore LL scheduler #10949

Merged

lyakh mentioned this pull request Jul 10, 2026

subset of #10945 #10984

Merged

lyakh force-pushed the llext-ull branch 2 times, most recently from a462fcf to 9f6770d Compare July 14, 2026 09:58

This was referenced Jul 14, 2026

ipc: add support for userspace IPC handling #10994

Merged

userspace: a subset of #10945 that can be applied already now #11017

Merged

lyakh force-pushed the llext-ull branch from 4408350 to c20c623 Compare August 5, 2026 07:27

lyakh mentioned this pull request Aug 5, 2026

[Userspace LL] scattered commits from #10945 without dependencies #11065

Merged

This was referenced Aug 12, 2026

[Userspace LL] app: overlays: ptl: ll_userspace: disable DP scheduler #11085

Merged

[DRAFT/TESTING] userspace LL/audio test PR (latest ver: V43) #10558

Open

lyakh force-pushed the llext-ull branch from c20c623 to 4aebe2b Compare August 20, 2026 13:21

lyakh force-pushed the llext-ull branch from 4aebe2b to c2041bb Compare August 20, 2026 13:34

This was referenced Aug 20, 2026

[userspace LL] vregion related syscalls #11108

Merged

schedule: add a function for finding userspace scherulers #11109

Merged

scattered fixes and cosmetic improvements #11110

Merged

[userspace LL] LLEXT part of #10945 #11112

Merged

lyakh force-pushed the llext-ull branch from c2041bb to 389f5a5 Compare August 27, 2026 13:39

lyakh added the DNM Do Not Merge tag label Aug 27, 2026

lyakh marked this pull request as ready for review August 27, 2026 13:40

Copilot AI lite review requested due to automatic review settings August 27, 2026 13:40

lyakh requested review from LaurentiuM1234, abonislawski, dbaluta, iuliana-prodan, kv2019i, marcinszkudlinski and pblaszko as code owners August 27, 2026 13:40

lyakh force-pushed the llext-ull branch from e62b4cd to 69f09d7 Compare August 28, 2026 08:11

This was referenced Aug 28, 2026

[userspace LL] IPC part of #10945 #11137

Merged

[userspace LL] Scheduling part of #10945 #11138

Merged

[userspace LL] library manager commits from #10945 #11139

Closed

lyakh force-pushed the llext-ull branch from 69f09d7 to 524d832 Compare August 28, 2026 15:33

lyakh force-pushed the llext-ull branch from 34ad06a to 4f17b68 Compare September 2, 2026 08:04

lyakh changed the title ~~[DNM][Do Not Review][WiP][PoC] userspace LL scheduling: LLEXT & multicore~~ [userspace LL] scheduling: LLEXT, DP & multicore Sep 2, 2026

lyakh requested a lite review from Copilot September 2, 2026 08:26

Copilot started reviewing on behalf of lyakh September 2, 2026 08:28 View session

Copilot AI reviewed Sep 2, 2026

View reviewed changes

Comment thread src/ipc/ipc-common.c

Comment thread src/library_manager/llext_manager.c

kv2019i reviewed Sep 2, 2026

View reviewed changes

lyakh force-pushed the llext-ull branch from 4f17b68 to e9114c6 Compare September 3, 2026 12:09

lyakh force-pushed the llext-ull branch from e9114c6 to b52af6c Compare September 3, 2026 12:20

kv2019i approved these changes Sep 3, 2026

View reviewed changes

lyakh added 8 commits September 4, 2026 10:27

llext: with userspace let LL thread access DP modules too

d41a47e

If LL runs in userspace, it needs access to loaded LLEXT modules, running in DP more too. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

schedule: ll: enable multicore userspace

70133bb

Make scheduling LL thread and synchronisation objects per-core and forward IPCs and scheduling events accordingly. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

schedule: dp: application: fix NULL check order

6486918

Checking pointer for non-NULL after dereferencing it makes no sense. Swap the order. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

schedue: dp: used cached core ID values

90eb976

Simplify code by using cached core ID values instead of querying the hardware or re-evaluating expressions. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

lyakh force-pushed the llext-ull branch from b52af6c to 6a70cdd Compare September 4, 2026 08:29

lyakh removed the DNM Do Not Merge tag label Sep 4, 2026

kv2019i merged commit 90687ac into thesofproject:main Sep 4, 2026
44 of 46 checks passed

lyakh deleted the llext-ull branch September 4, 2026 10:38

Uh oh!

Conversation

lyakh commented Jun 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

lyakh commented Aug 20, 2026

Uh oh!

lyakh commented Aug 27, 2026

Uh oh!

intel-sofci commented Sep 1, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

PR 10945: test results

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🟡 Changes recommended

Suppressed comments (1)

Uh oh!

Uh oh!

Uh oh!

kv2019i left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

kv2019i Sep 2, 2026

Choose a reason for hiding this comment

Uh oh!

lyakh Sep 3, 2026

Choose a reason for hiding this comment

Uh oh!

kv2019i Sep 3, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

kv2019i Sep 2, 2026

Choose a reason for hiding this comment

Uh oh!

lyakh commented Sep 3, 2026

Uh oh!

lyakh commented Sep 3, 2026

Uh oh!

kv2019i commented Sep 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

lyakh commented Jun 18, 2026 •

edited

Loading

intel-sofci commented Sep 1, 2026 •

edited

Loading

kv2019i commented Sep 3, 2026 •

edited

Loading