Page MenuHomeFreeBSD

mmacy (Matthew Macy)
User

Projects

User Details

User Since
May 4 2018, 7:26 PM (32 w, 2 d)

Recent Activity

Fri, Dec 14

mmacy updated the diff for D18520: Update opencrypto for ZFS crypto.

fix build

Fri, Dec 14, 1:53 AM

Thu, Dec 13

mmacy updated the diff for D18520: Update opencrypto for ZFS crypto.

replace GMAC_DIGEST_LEN with CCM_CBC_MAX_DIGEST_LEN

Thu, Dec 13, 5:41 AM
mmacy added a comment to D18520: Update opencrypto for ZFS crypto.

I think I've incorporated all feedback but the MAC len name change.

Thu, Dec 13, 5:35 AM
mmacy updated the diff for D18520: Update opencrypto for ZFS crypto.
  • incorporate two missing bits of feedback
Thu, Dec 13, 5:34 AM
mmacy added inline comments to D18520: Update opencrypto for ZFS crypto.
Thu, Dec 13, 5:26 AM
mmacy updated the diff for D18520: Update opencrypto for ZFS crypto.
  • rebase post D18522 commit
  • incorporate some of cem's feedback
Thu, Dec 13, 5:26 AM
mmacy committed rS342024: Generalize AES iov optimization.
Generalize AES iov optimization
Thu, Dec 13, 4:43 AM
mmacy closed D18522: AES iov optimization.
Thu, Dec 13, 4:43 AM

Wed, Dec 12

mmacy added a reviewer for D18522: AES iov optimization: cem.
Wed, Dec 12, 5:07 AM

Tue, Dec 11

mmacy updated the diff for D18520: Update opencrypto for ZFS crypto.
  • fix numbering
  • pull in cryptocheck change
Tue, Dec 11, 11:58 PM
mmacy added reviewers for D18520: Update opencrypto for ZFS crypto: allanjude, cperciva.
Tue, Dec 11, 11:06 PM
mmacy created D18520: Update opencrypto for ZFS crypto.
Tue, Dec 11, 10:59 PM

Tue, Nov 27

mmacy added a comment to D17862: emulators/qemu-powernv: rename emulators/qemu-ppcnv and update to POWER9.

@krion @timur poke

Tue, Nov 27, 4:18 AM

Sun, Nov 25

mmacy added a comment to D18327: graphics/drm-next-kmod: Add hack to help make AMD drivers usable.

it's a last resort mechanism, as opposed to the regular PID-controller pagedaemons

It should run regularly (with a 10s period) whenever the pagedaemon is attempting to reclaim pages.

Is there a specific application you're using to trigger this problem? I've never seen a large number of pages be wired by the TTM; maybe there's a GEM object leak? Are you using amdgpu or radeonkms?

Sun, Nov 25, 11:00 PM

Wed, Nov 21

mmacy added a reviewer for D18041: LKPI updates for drm-v4.18: bwidawsk.
Wed, Nov 21, 9:21 PM
mmacy updated the diff for D18041: LKPI updates for drm-v4.18.
  • respond to feedback
  • further changes for 4.18
Wed, Nov 21, 7:44 PM
mmacy added a reviewer for D17982: Add support for newer DWARF fields to ldscript: dim.
Wed, Nov 21, 6:20 AM
mmacy committed rS340718: hwpmc: new counters and bug fix MFCs.
hwpmc: new counters and bug fix MFCs
Wed, Nov 21, 1:24 AM

Mon, Nov 19

mmacy accepted D18052: Summary: linuxkpi: Add some basic swap functions These are used by kms-drm to determine various heuristics related to low memory conditions..

+1 I was just observing the need for this last night.

Mon, Nov 19, 11:56 PM
mmacy accepted D18026: Enable evdev on ppc64.

LGTM

Mon, Nov 19, 6:52 AM · PowerPC
mmacy created D18041: LKPI updates for drm-v4.18.
Mon, Nov 19, 2:11 AM

Nov 14 2018

mmacy committed rS340420: epoch(9) revert r340097 - no longer a need for multiple sections per cpu.
epoch(9) revert r340097 - no longer a need for multiple sections per cpu
Nov 14 2018, 12:12 AM

Nov 13 2018

mmacy created D17982: Add support for newer DWARF fields to ldscript.
Nov 13 2018, 11:27 PM
mmacy committed rS340414: Add ZFS to amd64 NOTES to catch future breakage of static linking.
Add ZFS to amd64 NOTES to catch future breakage of static linking
Nov 13 2018, 11:09 PM
mmacy added a comment to D17860: <<Two NVMe Bootup hang>.

Why isn’t this hit elsewhere? Are you compiling without EARLY_AP ?

Nov 13 2018, 9:36 PM
mmacy accepted D17901: Use busdma unconditionally in iflib.
Nov 13 2018, 9:25 PM

Nov 9 2018

mmacy updated the summary of D17933: add NUMA support to powerpc.
Nov 9 2018, 9:53 PM
mmacy created D17933: add NUMA support to powerpc.
Nov 9 2018, 9:46 PM

Nov 6 2018

mmacy abandoned D17011: hwpmc: Refactor sample ring buffer handling.

Fixed

Nov 6 2018, 8:36 PM

Nov 5 2018

mmacy committed rS340149: hwpmc: limit wait for user callchain collection to 1 tick.
hwpmc: limit wait for user callchain collection to 1 tick
Nov 5 2018, 8:11 AM
mmacy committed rS340148: e1000: Don't use 9k jumbo clusters.
e1000: Don't use 9k jumbo clusters
Nov 5 2018, 6:59 AM
mmacy closed D16534: e1000: Don't use 9k jumbo clusters.
Nov 5 2018, 6:59 AM

Nov 4 2018

mmacy committed rS340145: Backport of r338074 - generalize uart_bus_probe and add SNPS support to x86.
Backport of r338074 - generalize uart_bus_probe and add SNPS support to x86
Nov 4 2018, 11:29 PM
mmacy closed D17381: Backport of rs338074 for stable/11 branch.
Nov 4 2018, 11:29 PM
mmacy added a comment to D17381: Backport of rs338074 for stable/11 branch.

Odd. It applied cleanly this time. I'll commit after a tinderbox run.

Nov 4 2018, 6:31 AM
mmacy committed rS340108: Add aditional counter descriptions to AMD 0x17.
Add aditional counter descriptions to AMD 0x17
Nov 4 2018, 6:25 AM
mmacy closed D17401: Add new counter descriptions to AMD 0x17.
Nov 4 2018, 6:24 AM
mmacy accepted D17513: Add missing performance optimization of AVL tree comparator functions.
Nov 4 2018, 5:09 AM
mmacy accepted D17617: Fix for use-after-free in if_vlan.c.
Nov 4 2018, 5:06 AM

Nov 3 2018

mmacy reopened D17011: hwpmc: Refactor sample ring buffer handling.

After this change sampling under load is truly terrible - rejecting 99+% of samples:

+ pmcstat -S unhalted_core_cycles -O ppid.pmcstat sleep 10
+ pmcstat -R ppid.pmcstat -z100 -G ppid.stacks
CONVERSION STATISTICS:
 #exec/elf                                1
 #samples/total                           566445
 #callchain/dubious-frames                565995
Nov 3 2018, 7:44 AM
mmacy updated subscribers of D17011: hwpmc: Refactor sample ring buffer handling.
Nov 3 2018, 7:43 AM
mmacy committed rS340097: Convert epoch to read / write records per cpu.
Convert epoch to read / write records per cpu
Nov 3 2018, 5:18 AM
mmacy committed rS340097: Convert epoch to read / write records per cpu.
Convert epoch to read / write records per cpu
Nov 3 2018, 5:15 AM

Nov 2 2018

mmacy requested changes to D17787: While debugging some epoch related races at Netflix, we discoveredfew non fundamental, but annoying issues with epoch.First, the inlining makes it difficult to profile and trace epoch.At the same time, inlining doesn't effectively happens. In....

This needs to be multiple reviews.

Nov 2 2018, 8:39 AM
mmacy added a comment to D17787: While debugging some epoch related races at Netflix, we discoveredfew non fundamental, but annoying issues with epoch.First, the inlining makes it difficult to profile and trace epoch.At the same time, inlining doesn't effectively happens. In....

Please separate out the de-inlining in to a separate review. We can commit that as the major chunk and then the other changes will be easier to review.

Nov 2 2018, 8:38 AM
mmacy added a comment to D17787: While debugging some epoch related races at Netflix, we discoveredfew non fundamental, but annoying issues with epoch.First, the inlining makes it difficult to profile and trace epoch.At the same time, inlining doesn't effectively happens. In....

sorry didn't mean to change the title - just add markj

Nov 2 2018, 8:33 AM
mmacy retitled D17787: While debugging some epoch related races at Netflix, we discoveredfew non fundamental, but annoying issues with epoch.First, the inlining makes it difficult to profile and trace epoch.At the same time, inlining doesn't effectively happens. In... from While debugging some epoch related races at Netflix, we discovered few non fundamental, but annoying issues with epoch. First, the inlining makes it difficult to profile and trace epoch. At the same time, inlining doesn't effectively happens. In... to While debugging some epoch related races at Netflix, we discoveredfew non fundamental, but annoying issues with epoch.First, the inlining makes it difficult to profile and trace epoch.At the same time, inlining doesn't effectively happens. In....
Nov 2 2018, 8:13 AM

Oct 29 2018

mmacy added a comment to D17401: Add new counter descriptions to AMD 0x17.

Do you need this in for the 12 release? I can do that, it just means some more hoops for me jump through.

I am unable to commit (svn commit). (https://wiki.freebsd.org/Phabricator)

Do I have to commit or it will be automatically do ?? do I miss any steps ??

svn-commit.tmp: 36 lines, 1596 characters.
svn: E170001: Commit failed (details follow):
svn: E170001: Authorization failed
svn: E170001: Your commit message was left in a temporary file:
svn: E170001: '/usr/home/amar/freebsd_svn/freebsd_head/svn-commit.tmp'

Someone with a commit bit needs to do it

Hi Macy can you help me to commit this changes..

Oct 29 2018, 9:11 PM
mmacy added a comment to D17601: ppc64: limited 32-bit DMA address range.

Also worth looking at:
https://lists.ozlabs.org/pipermail/linuxppc-dev/2018-June/174985.html

Oct 29 2018, 6:39 PM
mmacy added a comment to D17601: ppc64: limited 32-bit DMA address range.

I just rebased to today's HEAD from that of the 6th. This change makes my system unbootable. I haven't delved in to it, but my guess is I don't have enough contiguous memory below the limit (my system has 512GB)

Oct 29 2018, 2:03 AM

Oct 26 2018

mmacy committed rS339771: fix up more issues introduced by failing to have run TB.
fix up more issues introduced by failing to have run TB
Oct 26 2018, 8:52 AM
mmacy committed rS339769: fix i386 breakage caused by r339767.
fix i386 breakage caused by r339767
Oct 26 2018, 6:13 AM
mmacy committed rS339767: hwpmc: Enable hwpmc support for AMD Family 17H devices.
hwpmc: Enable hwpmc support for AMD Family 17H devices
Oct 26 2018, 5:13 AM
mmacy added a comment to D17381: Backport of rs338074 for stable/11 branch.

@rajfbsd_gmail.com I can't actually apply this to a release branch (11.2) I can only commit to stable/11 (from which 11.3 will come) where it looks like parts of this have already been MFC'd. The patch doesn't actually apply cleanly there. Could you please update it to apply to the stable/11 branch?

Oct 26 2018, 2:19 AM
mmacy added a comment to D17682: Fix get_maxfds() in jevents.

Yes, the current set of patches required (for macos and linux) can be seen here: https://github.com/arichardson/freebsd/tree/crossbuild-aug2018?files=1

Oct 26 2018, 12:37 AM

Oct 25 2018

mmacy added a comment to D16894: Put building of drm and drm2 modules behind options and Add big, nasty abandonware tags to this code..

should we abandon?

Oct 25 2018, 7:45 PM
mmacy abandoned D16876: Expand min/max_offset macros in vm_map.h to reduce pollution.

Already effectuated in HEAD

Oct 25 2018, 7:44 PM
mmacy commandeered D16876: Expand min/max_offset macros in vm_map.h to reduce pollution.
Oct 25 2018, 7:44 PM
mmacy accepted D17381: Backport of rs338074 for stable/11 branch.

It looks fine to me. I'll commit it as soon as I can.

Oct 25 2018, 7:35 PM
mmacy added a comment to D17593: in[6]_pcblookup_hash_locked() and callers locking after epoch(9).
In D17593#375500, @bz wrote:

Is this in response to an observed bug? This is adding a lot of complexity when we can simply validate the result

The complexity mostly comes from the fact that we can (and do) now properly lock in the inp when going over the lists (which we couldn't before due to lock order) and with that are less prone to races. Hence this is not just about validating the inp is not FREED but one patch without the other half would make it worse. I can break them up into two if that makes it easier to review/progress.

Oct 25 2018, 7:20 PM
mmacy added a comment to D17598: Properly handle case when system is out of network interface numbers.

@hselasky this is getting super complicated - how much can we avoid by replacing the epoch_wait()+free() with an epoch_call()?

Oct 25 2018, 7:18 PM
mmacy added a comment to D17599: Fix for double free when deleting entries from epoch managed lists.

@hselasky have you had pho@ test this change?

Oct 25 2018, 7:04 PM
mmacy accepted D17682: Fix get_maxfds() in jevents.

LGTM. Are you cross-building from OSX?

Oct 25 2018, 6:51 PM

Oct 18 2018

mmacy accepted D17604: Maintain qualifiers on (struct thread_lite) members.
Oct 18 2018, 2:00 AM

Oct 17 2018

mmacy updated subscribers of D17593: in[6]_pcblookup_hash_locked() and callers locking after epoch(9).
Oct 17 2018, 5:57 AM
mmacy added a comment to D17593: in[6]_pcblookup_hash_locked() and callers locking after epoch(9).

Is this in response to an observed bug? This is adding a lot of complexity when we can simply validate the result

Oct 17 2018, 5:54 AM

Oct 12 2018

mmacy accepted D17540: In udp_input() validate inp before use after epoch(9)..
Oct 12 2018, 10:14 PM
mmacy accepted D17496: Fix deadlock when destroying VLANs.
Oct 12 2018, 10:12 PM

Oct 11 2018

mmacy added a comment to D17401: Add new counter descriptions to AMD 0x17.

Do you need this in for the 12 release? I can do that, it just means some more hoops for me jump through.

I am unable to commit (svn commit). (https://wiki.freebsd.org/Phabricator)

Do I have to commit or it will be automatically do ?? do I miss any steps ??

svn-commit.tmp: 36 lines, 1596 characters.
svn: E170001: Commit failed (details follow):
svn: E170001: Authorization failed
svn: E170001: Your commit message was left in a temporary file:
svn: E170001: '/usr/home/amar/freebsd_svn/freebsd_head/svn-commit.tmp'

Oct 11 2018, 5:23 AM

Oct 10 2018

mmacy accepted D17401: Add new counter descriptions to AMD 0x17.

Do you need this in for the 12 release? I can do that, it just means some more hoops for me jump through.

Oct 10 2018, 10:52 PM
mmacy added a comment to D17503: Run epoch calls sooner and more reliably.
In D17503#373585, @jtl wrote:
In D17503#373529, @jtl wrote:

See previous note in D17492 regarding the assumption that write-side functions are *not* called with-in a read-side critical. Does FreeBSD require that this is permitted?

See my comments in that review.

Regardless of whether that is true, it seems like this change would still be safe. Do you agree?

Is it expected by the kernel that ck_epoch write-side functions are safe to call on records that are already in a read-side critical section?
Helping me understand the intended use-case will allow me to answer accurately. There is further validation needed beyond poll.

See related comments in D17492. In short, the kernel does expect to be able to call ck_epoch_call() and ck_epoch_poll_deferred() on records that are already in a read-side critical section.

Oct 10 2018, 7:54 PM
mmacy added a comment to D17503: Run epoch calls sooner and more reliably.
In D17503#373529, @jtl wrote:

See previous note in D17492 regarding the assumption that write-side functions are *not* called with-in a read-side critical. Does FreeBSD require that this is permitted?

See my comments in that review.

Regardless of whether that is true, it seems like this change would still be safe. Do you agree?

Is it expected by the kernel that ck_epoch write-side functions are safe to call on records that are already in a read-side critical section?
Helping me understand the intended use-case will allow me to answer accurately. There is further validation needed beyond poll.

Oct 10 2018, 7:40 PM

Oct 7 2018

mmacy accepted D17450: Fix three epoch leaks.
Oct 7 2018, 5:55 PM

Oct 5 2018

mmacy accepted D17419: Use malloc_domainset(9) instead of malloc_domain(M_WAITOK)..

As I pointed out on IRC, this problem is not specific to hwpmc. Nonetheless it does fix the issue here.

Oct 5 2018, 7:58 PM
mmacy committed rS339188: hwpmc: Refactor sample ring buffer handling to fix races.
hwpmc: Refactor sample ring buffer handling to fix races
Oct 5 2018, 5:56 AM
mmacy closed D17011: hwpmc: Refactor sample ring buffer handling.
Oct 5 2018, 5:56 AM
mmacy committed rS339187: eliminate locking surrounding ui_vmsize and swap reserve by using atomics.
eliminate locking surrounding ui_vmsize and swap reserve by using atomics
Oct 5 2018, 5:51 AM
mmacy closed D16273: eliminate locking surrounding ui_vmsize and swap reserve by using atomics.
Oct 5 2018, 5:51 AM
mmacy added a comment to D17427: <<hwpmc> Add library and kernel support for AMD Family 17h new counters L3 Cache and Data Fabric.

Is there a reason you didn't just update the diff on the existing review?

Oct 5 2018, 5:48 AM
mmacy added reviewers for D17426: Speed up devel/gdb startup on FreeBSD: jhb, jmd.
Oct 5 2018, 12:25 AM

Oct 4 2018

mmacy added inline comments to D17401: Add new counter descriptions to AMD 0x17.
Oct 4 2018, 7:58 PM
mmacy retitled D17401: Add new counter descriptions to AMD 0x17 from <<hwpmc> to Add new counter descriptions to AMD 0x17.
Oct 4 2018, 7:52 PM
mmacy updated the diff for D16273: eliminate locking surrounding ui_vmsize and swap reserve by using atomics.
  • add #ifdef RACCT
  • fix typo
Oct 4 2018, 4:51 AM

Oct 3 2018

mmacy updated the diff for D16273: eliminate locking surrounding ui_vmsize and swap reserve by using atomics.

incorporate @alc feedback

Oct 3 2018, 11:05 PM
mmacy updated the summary of D17011: hwpmc: Refactor sample ring buffer handling.
Oct 3 2018, 10:12 PM
mmacy added a comment to D17381: Backport of rs338074 for stable/11 branch.

@imp anyone else we should add as a reviewer?

Oct 3 2018, 10:04 PM
mmacy added a reviewer for D17381: Backport of rs338074 for stable/11 branch: imp.
Oct 3 2018, 10:03 PM
mmacy updated the diff for D17011: hwpmc: Refactor sample ring buffer handling.

incorporate feedback from @markj

Oct 3 2018, 9:29 PM
mmacy added inline comments to D17011: hwpmc: Refactor sample ring buffer handling.
Oct 3 2018, 9:25 PM
mmacy added inline comments to D17011: hwpmc: Refactor sample ring buffer handling.
Oct 3 2018, 9:07 PM
mmacy added a comment to D17011: hwpmc: Refactor sample ring buffer handling.

I’ll update the patch with the other feedback.

Oct 3 2018, 8:01 PM
mmacy added a comment to D16273: eliminate locking surrounding ui_vmsize and swap reserve by using atomics.

@kib any further comments?

Oct 3 2018, 6:32 PM
mmacy added a comment to D17011: hwpmc: Refactor sample ring buffer handling.

@kib @markj Any further comments before I send this to RE?

Oct 3 2018, 6:32 PM

Oct 2 2018

mmacy added a comment to D17011: hwpmc: Refactor sample ring buffer handling.
In D17011#363656, @kib wrote:

Can you add an explanation what exactly changes intend to fix, and how they achieve robustness improvements.

Oct 2 2018, 8:49 PM
mmacy added inline comments to D17011: hwpmc: Refactor sample ring buffer handling.
Oct 2 2018, 8:43 PM
mmacy updated the diff for D17011: hwpmc: Refactor sample ring buffer handling.

Fix patch

Oct 2 2018, 8:41 PM
mmacy updated the diff for D16273: eliminate locking surrounding ui_vmsize and swap reserve by using atomics.
  • use atop/ptoa macros
Oct 2 2018, 8:12 PM
mmacy added inline comments to D16273: eliminate locking surrounding ui_vmsize and swap reserve by using atomics.
Oct 2 2018, 8:10 PM
mmacy updated the diff for D17011: hwpmc: Refactor sample ring buffer handling.

remove invalid assert

Oct 2 2018, 8:03 PM
mmacy updated the diff for D17011: hwpmc: Refactor sample ring buffer handling.
  • remove unused force parameter
Oct 2 2018, 5:42 AM