Page MenuHomeFreeBSD

rlibby (Ryan Libby)
User

Projects

User Details

User Since
Nov 12 2015, 8:14 PM (218 w, 1 d)

Recent Activity

Yesterday

rlibby added inline comments to D23238: [uma-multipage 2/3] uma: convert mbuf_jumbo_alloc & pmap_pdpt_alloc to UMA_ZONE_CONTIG.
Fri, Jan 17, 8:54 PM
rlibby added a comment to D23238: [uma-multipage 2/3] uma: convert mbuf_jumbo_alloc & pmap_pdpt_alloc to UMA_ZONE_CONTIG.

Yes, that's right. In fact, the smaller mbufs are actually still okay with multipage slabs due to their power-of-2 size, but I thought it was better to make the constraint explicit.

Fri, Jan 17, 8:08 PM
rlibby added inline comments to D23239: [uma-multipage 3/3] uma: grow slabs to enforce minimum memory efficiency.
Fri, Jan 17, 6:16 PM
rlibby added inline comments to D23237: [uma-multipage 1/3] uma: add UMA_ZONE_CONTIG, and a default contig_alloc.
Fri, Jan 17, 6:08 PM
rlibby updated subscribers of D23237: [uma-multipage 1/3] uma: add UMA_ZONE_CONTIG, and a default contig_alloc.
Fri, Jan 17, 5:56 PM
rlibby created D23239: [uma-multipage 3/3] uma: grow slabs to enforce minimum memory efficiency.
Fri, Jan 17, 5:53 PM
rlibby created D23238: [uma-multipage 2/3] uma: convert mbuf_jumbo_alloc & pmap_pdpt_alloc to UMA_ZONE_CONTIG.
Fri, Jan 17, 5:47 PM
rlibby created D23237: [uma-multipage 1/3] uma: add UMA_ZONE_CONTIG, and a default contig_alloc.
Fri, Jan 17, 5:44 PM
rlibby created D23223: auditd_stop: wait_for_pids instead of sleeping.
Fri, Jan 17, 7:08 AM

Tue, Jan 14

rlibby committed rS356717: uma: split slabzone into two sizes.
uma: split slabzone into two sizes
Tue, Jan 14, 2:14 AM
rlibby closed D23149: uma: split slabzone into two sizes.
Tue, Jan 14, 2:14 AM
rlibby committed rS356716: malloc: remove assumptions about MINALLOCSIZE.
malloc: remove assumptions about MINALLOCSIZE
Tue, Jan 14, 2:14 AM
rlibby closed D23148: uma: fixup some ktr messages.
Tue, Jan 14, 2:14 AM
rlibby committed rS356715: uma: fixup some ktr messages.
uma: fixup some ktr messages
Tue, Jan 14, 2:14 AM

Mon, Jan 13

rlibby added inline comments to D23149: uma: split slabzone into two sizes.
Mon, Jan 13, 6:26 PM
rlibby added inline comments to D23149: uma: split slabzone into two sizes.
Mon, Jan 13, 5:19 PM
rlibby updated the diff for D23149: uma: split slabzone into two sizes.

markj feedback: style(9)

Mon, Jan 13, 5:13 PM
rlibby added inline comments to D23149: uma: split slabzone into two sizes.
Mon, Jan 13, 8:56 AM
rlibby created D23149: uma: split slabzone into two sizes.
Mon, Jan 13, 8:53 AM
rlibby updated the diff for D23148: uma: fixup some ktr messages.

missed one, and messed one up

Mon, Jan 13, 1:36 AM
rlibby updated the diff for D23148: uma: fixup some ktr messages.

markj feedback, don't print threads

Mon, Jan 13, 12:34 AM

Sun, Jan 12

rlibby added inline comments to D23148: uma: fixup some ktr messages.
Sun, Jan 12, 8:49 PM
rlibby created D23148: uma: fixup some ktr messages.
Sun, Jan 12, 8:04 PM

Thu, Jan 9

rlibby added inline comments to D23114: Document two new conventions:.
Thu, Jan 9, 11:07 PM
rlibby added a comment to D22963: Expand generic subword atomic primitives.
In D22963#506510, @cem wrote:

I don't think expecting misaligned atomics to work is reasonable. We could KASSERT alignment? Asking for 2-byte alignment isn't asking much.

Thu, Jan 9, 7:13 PM
rlibby added inline comments to D22963: Expand generic subword atomic primitives.
Thu, Jan 9, 5:16 PM
rlibby added inline comments to D22963: Expand generic subword atomic primitives.
Thu, Jan 9, 10:11 AM
rlibby added a comment to D22963: Expand generic subword atomic primitives.

I only looked at _atomic_subword.h.

Thu, Jan 9, 9:33 AM
rlibby committed rS356535: uma: unify layout paths and improve efficiency.
uma: unify layout paths and improve efficiency
Thu, Jan 9, 2:03 AM
rlibby closed D23048: uma: unify layout paths and improve efficiency.
Thu, Jan 9, 2:03 AM
rlibby closed D23016: uma: reorganize flags.
Thu, Jan 9, 2:03 AM
rlibby committed rS356534: uma: reorganize flags.
uma: reorganize flags
Thu, Jan 9, 2:03 AM

Wed, Jan 8

rlibby accepted D23095: malloc_domain(): return the allocation size to the caller..

Looks good. Did this fix the bus-sc accounting?

Wed, Jan 8, 10:23 PM
rlibby accepted D23066: Don't bother freeing UMA zones after the shutdown process starts..
Wed, Jan 8, 5:40 PM
rlibby added inline comments to D23048: uma: unify layout paths and improve efficiency.
Wed, Jan 8, 9:01 AM
rlibby updated the diff for D23048: uma: unify layout paths and improve efficiency.

jeff feedback, s/ipers_nohdr/ipers_offpage/ too

Wed, Jan 8, 8:59 AM
rlibby updated the diff for D23048: uma: unify layout paths and improve efficiency.

forgot jeff feedback s/eff_nohdr/eff_offpage/

Wed, Jan 8, 8:43 AM
rlibby added inline comments to D23048: uma: unify layout paths and improve efficiency.
Wed, Jan 8, 8:33 AM
rlibby updated the diff for D23048: uma: unify layout paths and improve efficiency.

jeff & markj feedback

Wed, Jan 8, 8:32 AM
rlibby added a comment to D23066: Don't bother freeing UMA zones after the shutdown process starts..

Code looks good.

Wed, Jan 8, 5:31 AM

Mon, Jan 6

rlibby added a comment to D23048: uma: unify layout paths and improve efficiency.

@jeff Will do for the rest of the comments. A couple clarifying questions.

Mon, Jan 6, 8:05 PM
rlibby added inline comments to D23048: uma: unify layout paths and improve efficiency.
Mon, Jan 6, 5:20 PM
rlibby added a comment to D23048: uma: unify layout paths and improve efficiency.

Jeff asked that the old code be left in place for posting the diff, for readability of the diff. The #ifdef 0 code will be deleted before commit.

Mon, Jan 6, 1:02 AM
rlibby created D23048: uma: unify layout paths and improve efficiency.
Mon, Jan 6, 1:00 AM
rlibby accepted D23046: Fix zone and keg alignment as well as boot pages calculations on numa..
Mon, Jan 6, 12:10 AM

Sat, Jan 4

rlibby added a comment to D22830: (umaperf 6/7) Sort crossdomain free buckets into domain correct buckets before returning to the system..

Latest revision LGTM.

Sat, Jan 4, 7:38 PM

Fri, Jan 3

rlibby added inline comments to D23016: uma: reorganize flags.
Fri, Jan 3, 7:24 PM
rlibby created D23016: uma: reorganize flags.
Fri, Jan 3, 1:27 AM

Mon, Dec 30

rlibby accepted D22964: rtld(1): Do booleans like C99.

This one blames to 1998, so I guess it was legit :)

Mon, Dec 30, 5:21 PM

Wed, Dec 25

rlibby added a comment to D22876: gcc9: quiet Waddress-of-packed-member for kernel build.

Here's what changing it to no-error for the clang build does (amd64):

Wed, Dec 25, 5:22 AM

Tue, Dec 24

rlibby added inline comments to D22830: (umaperf 6/7) Sort crossdomain free buckets into domain correct buckets before returning to the system..
Tue, Dec 24, 10:09 PM
rlibby added inline comments to D22830: (umaperf 6/7) Sort crossdomain free buckets into domain correct buckets before returning to the system..
Tue, Dec 24, 8:09 PM
rlibby accepted D22829: (umaperf 5/7) Use per-domain keg locks..

Looks good, after fixing up the keg_alloc_slab comment and the keg_dtor printf format.

Tue, Dec 24, 6:05 PM
rlibby accepted D22914: sleep(9), sleepqueue(9): const'ify wchan pointers.

Rebase LGTM.

Tue, Dec 24, 8:16 AM
rlibby added a comment to D22829: (umaperf 5/7) Use per-domain keg locks..

Looks good but I do have a question about how uma_zone_reserve() is supposed to work now.

Tue, Dec 24, 8:02 AM
rlibby accepted D22914: sleep(9), sleepqueue(9): const'ify wchan pointers.

Looks good. Recommend building at least INVARIANTS and non-INVARIANTS kernels before commit.

Tue, Dec 24, 2:08 AM

Mon, Dec 23

rlibby accepted D22827: (umaperf 3/7) atomic zone limits.

Fixups look good. I agree with leaving an XXX and coming back later to the problem of transitioning between no-limit and limit.

Mon, Dec 23, 9:33 PM
rlibby added inline comments to D22827: (umaperf 3/7) atomic zone limits.
Mon, Dec 23, 8:52 PM
rlibby added a comment to D22827: (umaperf 3/7) atomic zone limits.

Generally looking good, but I do have a few more comments.

Mon, Dec 23, 8:44 PM

Sun, Dec 22

rlibby accepted D22826: (umaperf 2/7) Store the zone flags and size in the uma per-cpu cache area..

Fixups look good.

Sun, Dec 22, 9:57 PM
rlibby accepted D22825: (umaperf 1/7) Save an extra cache line access for large buckets as well as a few branches..

Fixups look good.

Sun, Dec 22, 9:35 PM

Sat, Dec 21

rlibby added a comment to D22876: gcc9: quiet Waddress-of-packed-member for kernel build.

I can prepare a follow up. What I did here was exactly what we already do for clang. I will change both. And then I assume you'd have the same comment for the user build?

Sat, Dec 21, 3:00 AM
rlibby committed rS355976: googletest: pick from upstream: Don't allow signed/unsigned wchar_t in gcc 9….
googletest: pick from upstream: Don't allow signed/unsigned wchar_t in gcc 9…
Sat, Dec 21, 2:45 AM
rlibby committed rS355974: libdevdctl: g++9 avoid Wdeprecated-copy.
libdevdctl: g++9 avoid Wdeprecated-copy
Sat, Dec 21, 2:44 AM
rlibby committed rS355975: jemalloc: pick from upstream: Fix GCC-9.1 warning with macro GET_ARG_NUMERIC.
jemalloc: pick from upstream: Fix GCC-9.1 warning with macro GET_ARG_NUMERIC
Sat, Dec 21, 2:44 AM
rlibby closed D22887: libdevdctl: g++9 avoid Wdeprecated-copy.
Sat, Dec 21, 2:44 AM
rlibby committed rS355973: dtrace: avoid gcc9 Walloca-larger-than.
dtrace: avoid gcc9 Walloca-larger-than
Sat, Dec 21, 2:44 AM
rlibby closed D22880: dtrace: quiet gcc9 Walloca-larger-than.
Sat, Dec 21, 2:44 AM
rlibby committed rS355972: Declare packed struct ata_params as 2-byte-aligned.
Declare packed struct ata_params as 2-byte-aligned
Sat, Dec 21, 2:44 AM
rlibby closed D22888: Declare packed struct ata_params as 2-byte-aligned.
Sat, Dec 21, 2:44 AM
rlibby committed rS355971: gcc9: quiet Waddress-of-packed-member for user build.
gcc9: quiet Waddress-of-packed-member for user build
Sat, Dec 21, 2:44 AM
rlibby closed D22889: gcc9: quiet Waddress-of-packed-member for user build.
Sat, Dec 21, 2:44 AM
rlibby committed rS355970: gcc9: quiet Waddress-of-packed-member for kernel build.
gcc9: quiet Waddress-of-packed-member for kernel build
Sat, Dec 21, 2:43 AM
rlibby closed D22876: gcc9: quiet Waddress-of-packed-member for kernel build.
Sat, Dec 21, 2:43 AM
rlibby committed rS355969: gcc: quiet Wattribute for no_sanitize("address").
gcc: quiet Wattribute for no_sanitize("address")
Sat, Dec 21, 2:43 AM
rlibby closed D22875: gcc: quiet Wattribute for no_sanitize("address").
Sat, Dec 21, 2:43 AM

Fri, Dec 20

rlibby added a reviewer for D22889: gcc9: quiet Waddress-of-packed-member for user build: jhb.
Fri, Dec 20, 9:37 PM
rlibby added a comment to D22876: gcc9: quiet Waddress-of-packed-member for kernel build.
In D22876#501221, @jhb wrote:

Maybe include share/mk/bsd.sys.mk at the same time as userland needs it as well (at least it did for me)?

Fri, Dec 20, 9:36 PM
rlibby added a comment to D22888: Declare packed struct ata_params as 2-byte-aligned.
In D22888#501195, @imp wrote:

So it worked! Yippy! Of course I'd love you to commit this :)

Fri, Dec 20, 8:01 PM
rlibby retitled D22888: Declare packed struct ata_params as 2-byte-aligned from camcontrol: avoid gcc9 Waddress-of-packed-member to Declare packed struct ata_params as 2-byte-aligned.
Fri, Dec 20, 7:59 PM
rlibby updated the diff for D22888: Declare packed struct ata_params as 2-byte-aligned.

imp's suggestion

Fri, Dec 20, 7:57 PM
rlibby added a comment to D22888: Declare packed struct ata_params as 2-byte-aligned.

Ah. Yes. I'll check if that builds. Thanks.

Fri, Dec 20, 5:15 PM
rlibby added a comment to D22888: Declare packed struct ata_params as 2-byte-aligned.

A non-intrusive Makefile solution could be to lower camcontrol's WARNS to 3, either unconditionally or conditional on gcc.

Fri, Dec 20, 8:44 AM
rlibby added a comment to D22888: Declare packed struct ata_params as 2-byte-aligned.
In D22888#501055, @imp wrote:

It is actually properly aligned. Let's not do silly crap like this. Let's say it really is properly aligned.

Fri, Dec 20, 7:20 AM
rlibby updated the diff for D22880: dtrace: quiet gcc9 Walloca-larger-than.

Replace alloca

Fri, Dec 20, 7:18 AM
rlibby created D22889: gcc9: quiet Waddress-of-packed-member for user build.
Fri, Dec 20, 6:54 AM
rlibby created D22888: Declare packed struct ata_params as 2-byte-aligned.
Fri, Dec 20, 6:49 AM
rlibby created D22887: libdevdctl: g++9 avoid Wdeprecated-copy.
Fri, Dec 20, 6:30 AM

Thu, Dec 19

rlibby created D22880: dtrace: quiet gcc9 Walloca-larger-than.
Thu, Dec 19, 8:37 PM

Dec 19 2019

rlibby created D22876: gcc9: quiet Waddress-of-packed-member for kernel build.
Dec 19 2019, 9:50 AM
rlibby created D22875: gcc: quiet Wattribute for no_sanitize("address").
Dec 19 2019, 9:46 AM

Dec 17 2019

rlibby accepted D22826: (umaperf 2/7) Store the zone flags and size in the uma per-cpu cache area..

Looks good.

Dec 17 2019, 6:10 AM

Dec 16 2019

rlibby added inline comments to D22825: (umaperf 1/7) Save an extra cache line access for large buckets as well as a few branches..
Dec 16 2019, 7:25 AM
rlibby accepted D22825: (umaperf 1/7) Save an extra cache line access for large buckets as well as a few branches..

Looks good, just some nitpicks.

Dec 16 2019, 7:01 AM

Dec 14 2019

rlibby abandoned D22799: uma: fix user build.

I went ahead and committed this in r355746 along with re-application of D22759.

Dec 14 2019, 6:25 AM
rlibby committed rS355746: uma dbg: flexible size for slab debug bitset too.
uma dbg: flexible size for slab debug bitset too
Dec 14 2019, 5:22 AM

Dec 13 2019

rlibby created D22799: uma: fix user build.
Dec 13 2019, 8:11 PM
rlibby accepted D22797: Restore the reservation of boot pages for bucket zones..

The problem I was originally trying to solve was allowing embedded slabs on the bucket zones for D22768. This means the slab memory for the bucket zones must come from uma_small_alloc. It can't come from the startup pages because PHYS_TO_VM_PAGE doesn't work on the startup pages since they aren't in the page array (although, we could possibly change that, with some potential other cost, see the vm_page.c bit in D11784). So I added a check at keg creation time that only enables embedded slabs if the keg was created after BOOT_PAGEALLOC. That disabled embedding for bucket zones, which were being created before then.

Dec 13 2019, 5:48 PM
rlibby committed rS355711: Revert r355706 & r355710.
Revert r355706 & r355710
Dec 13 2019, 11:21 AM
rlibby committed rS355710: libmemstat: unbreak build.
libmemstat: unbreak build
Dec 13 2019, 10:34 AM