♟ markj

Simplify i915_gem_wire_page() and avoid unneeded page-busying.

Make vm_page_unwire(PQ_NONE) dequeue the page.
Deindent the vm_page_wire() herald comment to match that of vm_page_unwire().

MFC r322405, r322406:

Implement mmu_page_init for AIM platforms.

Include _bitset.h to get BITSET_DEFINE, used to define struct slabbits.

Widen uk_pgoff, the slab header offset field.

Fix a logic error in the item size calculation for internal UMA zones.

Add a O_CLOEXEC use missed in r323166.

LGTM, thanks Li-Wen.

Looks ok, but this code is pretty foreign to me. Could you point me to the doc(s) you consult for this stuff?

Provide an error message if KTR symbols in a vmcore cannot be resolved.

Address review comments; restore wire_count check in vm_pageout_cluster()

I think it'd be logical to move v_free_count as well, even if it doesn't make a difference now. The fields around it are read-mostly and generally are accessed only during memory shortages, but that might not be true in the future.

In D12281#255012, @mjg wrote:

The original fields are no longer named and compilation is ought to fail. As for code which wont be recompiled, your approach suffers the same problem by definition - since the layout of the struct has changed, old offsets don't fit. Short of proper ABI compatibility detection on module load there is nothing which can be done here.

Why can't you annotate the fields with __aligned(CACHE_LINE_SIZE), like:

(I'm holding off on this for a bit longer as I re-port netdump to HEAD. netdump is a bit different from a regular dump device since it doesn't have a fixed size, and I'm trying to avoid painting myself into a corner by assuming too much about how the dump device works.)

Speed up vm_page_array initialization.

Adress review feedback.

Also missing a makefile update.

This should go in cddl/lib/libdtrace so that it gets included automatically like the rest of the libraries. Otherwise it seems ok to me.

Avoid keeping a dangling pointer when the mappings array is resized.

Release rlibby from mentorship.

Use O_CLOEXEC when opening persistent handles in libdtrace.

MFC r322773-r322775:

Synchronize page laundering with pmap_extract_and_hold().

Set the bus number field when attaching a PCI device.

Add some miscellaneous definitions to support the DRM drivers.

Use an updated copy of the CDDL header boilerplate from illumos.

Add a regression test for r322773.

Fix an off-by-two in the llquantize() action parameter validation.

Define prefetch() only if it hasn't already been defined.

Add a couple of trivial headers to the LinuxKPI.

Looks ok to me, but I'm not very familiar with this code. I haven't hit any problems with it on a dual-package Xeon system.

Remove some unneeded subroutines for padding writes to dump devices.

Rename mkdumpheader() and group EKCD functions in kern_shutdown.c.

Factor out duplicated kernel dump code into dump_{start,finish}().

Add device resource management fields to struct device.

Add vm_page_alloc_after().

Eliminate bogus initialization.

Reference pages in PQ_ACTIVE upon unwiring them.
Remove wire_count checks from the page-clustering code.
Simplify the wire_count checking in vm_pageout_launder().
Hold off on adding a large comment to vm_page.h for now.

Reorder arguments to vm_page_alloc_after()
Use vm_radix_lookup_le() to initialize mpred

markj (Mark Johnston)
User

Projects

User Details

Recent Activity
View All

Today

Sun, Sep 17

Fri, Sep 15

Thu, Sep 14

Wed, Sep 13

Tue, Sep 12

Mon, Sep 11

Sun, Sep 10

Sat, Sep 9

Fri, Sep 8

Thu, Sep 7

Wed, Sep 6

Tue, Sep 5

Mon, Sep 4

Wed, Aug 30

Tue, Aug 29

Mon, Aug 28

Sat, Aug 26

Wed, Aug 23

Aug 22 2017

Aug 21 2017

Aug 20 2017

Aug 19 2017

Aug 18 2017

Aug 17 2017

Aug 16 2017

Aug 15 2017

Aug 11 2017

markj (Mark Johnston)User