Hey guys, I know that ifdefs suck, but still going to forward with checking this in. If you can suggest some change to ifdefs, I am open to that. E.g. keep them minimal, so there is no duplication between debug and !debug case. Or you may prefer to keep the maximal, so that code is more readable. I can do either, but I'd like to get forward with this change.

May 14 2018, 4:10 PM

May 1 2018

glebius added inline comments to D15199: Add possibility to disable or reduce amount of UMA debugging with INVARIANTS.

May 1 2018, 9:09 PM

Apr 27 2018

glebius added a comment to D15199: Add possibility to disable or reduce amount of UMA debugging with INVARIANTS.

In D15199#320705, @markj wrote:

I think all of these inline ifdefs make the code a lot harder to read. Could we instead introduce wrapper functions for the ctor and dtor and hide the logic there?

Apr 27 2018, 5:51 PM

Apr 26 2018

glebius added inline comments to D15199: Add possibility to disable or reduce amount of UMA debugging with INVARIANTS.

Apr 26 2018, 8:46 PM

Apr 25 2018

glebius created D15199: Add possibility to disable or reduce amount of UMA debugging with INVARIANTS.

Apr 25 2018, 5:03 PM

Apr 19 2018

glebius added a comment to D15112: In snmp_mibII module assume that virtual interfaces can work at maximum speed available in the system.

I did not do this in the past because it was not possible to add or increment 64-bit counters atomically. Did this change? Instead of providing bad values I decided to follow the words of the RFC.

Apr 19 2018, 9:32 PM

glebius updated the summary of D14980: pathconf(_PC_MEDIA) - an API to tell the media at a given path.

Apr 19 2018, 9:20 PM

Apr 16 2018

glebius committed rS332572: Merge r331871:.

Merge r331871:

Apr 16 2018, 3:07 PM

Apr 12 2018

glebius added a comment to D14980: pathconf(_PC_MEDIA) - an API to tell the media at a given path.

In D14980#316840, @mav wrote:

Is this API and device classification exist somewhere else, or being just designed? Differentiation between STRIPE, MIRROR and RAID sound somewhat odd to me, same as between SPINNING and CDROM. Besides it duplicates "GEOM::rotation_rate" attribute, we already have and use in some places (though obviously it is not passed through UFS).

Apr 12 2018, 9:18 PM

glebius accepted D15046: Make lagg creation more fault tolerant.

Apr 12 2018, 9:12 PM

Apr 11 2018

glebius updated the diff for D14980: pathconf(_PC_MEDIA) - an API to tell the media at a given path.

Support vnode backed md(4).
Bump disk version.

Apr 11 2018, 12:09 AM

glebius added inline comments to D14980: pathconf(_PC_MEDIA) - an API to tell the media at a given path.

Apr 11 2018, 12:01 AM

Apr 10 2018

glebius updated the diff for D14980: pathconf(_PC_MEDIA) - an API to tell the media at a given path.

Instead of using strcmp() augment struct disk with media info.

Apr 10 2018, 8:16 PM

Apr 5 2018

glebius added a comment to D14980: pathconf(_PC_MEDIA) - an API to tell the media at a given path.

# df
Filesystem        1K-blocks       Used     Avail Capacity  Mounted on
/dev/mirror/prim    3095920    1129672   1718576    40%    /
/dev/md5            5061628         24   5061604     0%    /var/svm
/dev/nda1p8      3585929296 3084809040 214245920    94%    /usr/local/www/2
/dev/mirror/logs   51581936    8159504  39295880    17%    /var/log
/dev/ada0p8       897802144  745401144  80576832    90%    /usr/local/www/a
/dev/da16p1      7691453560 6677369072 398768208    94%    /usr/local/www/12
# ./pathconf-media /usr/local/www/12
/usr/local/www/12=2001<SPINNING,SCSI>
# ./pathconf-media /usr/local/www/a
/usr/local/www/a=1002<SOLID,ATA>
# ./pathconf-media /var/log/
/var/log/=11002<SOLID,ATA,MIRROR>
# ./pathconf-media /usr/local/www/2
/usr/local/www/2=2<SOLID>
# ./pathconf-media /var/svm/
/var/svm/=8<MEMORY>
# ./pathconf-media /
/=10002<SOLID,MIRROR>

Apr 5 2018, 10:04 PM

glebius created D14980: pathconf(_PC_MEDIA) - an API to tell the media at a given path.

Apr 5 2018, 9:31 PM

Apr 4 2018

glebius added inline comments to D14893: VM page queue batching.

Apr 4 2018, 7:24 PM

Apr 3 2018

glebius committed rS331932: Indentation fixups. No functional changes..

Indentation fixups. No functional changes.

Apr 3 2018, 6:33 PM

Apr 2 2018

glebius committed rS331873: Use UMA_SLAB_SPACE macro. No functional change here..

Use UMA_SLAB_SPACE macro. No functional change here.

Apr 2 2018, 5:15 AM

glebius committed rS331872: In uma_startup_count() handle special case when zone will fit into.

In uma_startup_count() handle special case when zone will fit into

Apr 2 2018, 5:14 AM

glebius committed rS331871: Handle a special case when a slab can fit only one allocation,.

Handle a special case when a slab can fit only one allocation,

Apr 2 2018, 5:12 AM

Mar 22 2018

glebius committed rS331344: Redo r331328. We need to fix not only type but also format. While.

Redo r331328. We need to fix not only type but also format. While

Mar 22 2018, 5:26 AM

glebius committed rS331343: Fix LINT-NOINET build initializing local to false. This is.

Fix LINT-NOINET build initializing local to false. This is

Mar 22 2018, 5:08 AM

Mar 21 2018

glebius committed rS331328: Fix sysctl types broken in r329612..

Fix sysctl types broken in r329612.

Mar 21 2018, 11:21 PM

glebius committed rS331322: The net.inet.tcp.nolocaltimewait=1 optimization prevents local TCP connections.

The net.inet.tcp.nolocaltimewait=1 optimization prevents local TCP connections

Mar 21 2018, 8:59 PM

glebius closed D14697: Send a final ACK to local connections (version 2).

Mar 21 2018, 8:59 PM

Mar 20 2018

glebius committed rS331272: At this point iwmesg isn't initialized yet, so print pointer to lock.

At this point iwmesg isn't initialized yet, so print pointer to lock

Mar 20 2018, 10:05 PM

glebius committed rS331270: Fix whitespace issues in r330034. No functional changes..

Fix whitespace issues in r330034. No functional changes.

Mar 20 2018, 10:01 PM

Mar 14 2018

glebius added a comment to D14697: Send a final ACK to local connections (version 2).

My version doesn't allocate tcptw for local case, creating one on stack.

Mar 14 2018, 9:01 PM

glebius created D14697: Send a final ACK to local connections (version 2).

Mar 14 2018, 8:59 PM

Mar 13 2018

glebius committed rS330883: Fix typo that misteriously passes compilation..

Fix typo that misteriously passes compilation.

Mar 13 2018, 10:54 PM

Feb 9 2018

glebius committed rS329058: Fix boot_pages exhaustion on machines with many domains and cores, where.

Fix boot_pages exhaustion on machines with many domains and cores, where

Feb 9 2018, 4:45 AM

Feb 7 2018

glebius committed rS328982: Fix three miscalculations in amount of boot pages:.

Fix three miscalculations in amount of boot pages:

Feb 7 2018, 6:33 PM

Feb 6 2018

glebius committed rS328955: Use correct arithmetic to calculate how many pages we need for kegs.

Use correct arithmetic to calculate how many pages we need for kegs

Feb 6 2018, 10:13 PM

glebius committed rS328953: Improve DIAGNOSTIC printf. Report using a boot page every time regardless.

Improve DIAGNOSTIC printf. Report using a boot page every time regardless

Feb 6 2018, 10:08 PM

glebius committed rS328952: Fix boot_pages calculation for machines that don't have UMA_MD_SMALL_ALLOC..

Fix boot_pages calculation for machines that don't have UMA_MD_SMALL_ALLOC.

Feb 6 2018, 10:07 PM

glebius committed rS328916: Followup on r302393 by cperciva, improving calculation of boot pages required.

Followup on r302393 by cperciva, improving calculation of boot pages required

Feb 6 2018, 4:16 AM

glebius closed D14054: Better boot_pages autotuning..

Feb 6 2018, 4:16 AM

Jan 31 2018

glebius updated the summary of D14054: Better boot_pages autotuning..

Jan 31 2018, 6:55 PM

glebius updated the summary of D14054: Better boot_pages autotuning..

Jan 31 2018, 6:55 PM

glebius updated the diff for D14054: Better boot_pages autotuning..

A bigger approach to boot_pages management. The most important
change is that we actually can ask VM for pages much earlier
than we do now right after vm_page_startup(), thus do so.

Jan 31 2018, 6:51 PM

Jan 26 2018

glebius added inline comments to D14054: Better boot_pages autotuning..

Jan 26 2018, 11:02 PM

Jan 25 2018

glebius created D14054: Better boot_pages autotuning..

Jan 25 2018, 10:19 PM

Jan 5 2018

glebius committed rS327596: In sendfile_iodone() both pru_abort and sorele need to be executed.

In sendfile_iodone() both pru_abort and sorele need to be executed

Jan 5 2018, 8:21 PM

Dec 7 2017

glebius committed rS326673: Separate out send buffer autoscaling code into function, so that.

Separate out send buffer autoscaling code into function, so that

Dec 7 2017, 10:37 PM

Dec 6 2017

glebius committed rS326642: Garbage collect IFCAP_POLLING_NOCOUNT. It wasn't used since very.

Garbage collect IFCAP_POLLING_NOCOUNT. It wasn't used since very

Dec 6 2017, 11:03 PM

glebius committed rS326625: Fix crash with a dotless hostname..

Fix crash with a dotless hostname.

Dec 6 2017, 5:50 PM

glebius committed rS326608: Fix file missed in r326607..

Fix file missed in r326607.

Dec 6 2017, 12:45 AM

glebius committed rS326607: Reduce pollution via tmpfs.h..

Reduce pollution via tmpfs.h.

Dec 6 2017, 12:42 AM

Dec 5 2017

glebius committed rS326574: Generate fully RFC3164 compliant messages, with timestamp and hostname..

Generate fully RFC3164 compliant messages, with timestamp and hostname.

Dec 5 2017, 7:56 PM

glebius committed rS326573: When parsing remote messages, require them to have standard timestamp.

When parsing remote messages, require them to have standard timestamp

Dec 5 2017, 7:55 PM

Nov 28 2017

glebius committed rS326339: Revert r325558 as it is incorrect..

Revert r325558 as it is incorrect.

Nov 28 2017, 7:25 PM

glebius committed rS326338: Fix obvious NULL pointer dereference from r310350..

Fix obvious NULL pointer dereference from r310350.

Nov 28 2017, 7:18 PM

glebius committed rS326337: Revert r326103, as it appeared to be incorrect..

Revert r326103, as it appeared to be incorrect.

Nov 28 2017, 7:09 PM

Nov 22 2017

glebius committed rS326103: MFC r325558:.

MFC r325558:

Nov 22 2017, 9:20 PM

Nov 21 2017

glebius added a comment to D12914: Send a final ACK to local connections.

My fault, thanks for fixing.

Nov 21 2017, 10:53 PM

Nov 13 2017

glebius committed rS325782: Style r320614: don't initialize at declaration, new line after declarations,.

Style r320614: don't initialize at declaration, new line after declarations,

Nov 13 2017, 10:17 PM

Nov 8 2017

glebius committed rS325558: When parsing UDP messages skip optional hostname as described by.

When parsing UDP messages skip optional hostname as described by

Nov 8 2017, 4:46 PM

Oct 19 2017

glebius added a comment to D12404: Update tcpdump to 4.9.2..

D12405 looks good to me. Matches set of changed I had with 4.9.0. Is reduced actually, since tcpdump has incorporated some of our changes since.

Oct 19 2017, 10:15 PM

glebius accepted D12405: Diff showing stock tcpdump 4.9.2 vs FreeBSD 4.9.2.

Oct 19 2017, 10:12 PM

glebius added a comment to D12404: Update tcpdump to 4.9.2..

Hi! I think phabrickator is useless for such kinds of things. Did you follow instructions in vendor/tcpdump/FreeBSD-upgrade?

Oct 19 2017, 8:46 PM

glebius committed rS324767: Hide struct sockbuf from userland..

Hide struct sockbuf from userland.

Oct 19 2017, 6:04 PM

Oct 12 2017

glebius committed rS324569: Fix build after r324446..

Fix build after r324446.

Oct 12 2017, 9:27 PM

Oct 11 2017

glebius committed rS324539: Declare more TCP globals in tcp_var.h, so that alternative TCP stacks.

Declare more TCP globals in tcp_var.h, so that alternative TCP stacks

Oct 11 2017, 8:36 PM

Oct 10 2017

glebius requested changes to D12583: check so_error earlier in sosend_generic.

Oct 10 2017, 8:10 PM

glebius added a comment to D12583: check so_error earlier in sosend_generic.

SUS specifies ECONNRESET for write(2), but BSD systems never did that. SUS also specifies EPIPE, and doesn't provide any clue when ECONNRESET is preferred over EPIPE. They got very similar descriptions.

Oct 10 2017, 8:09 PM

glebius accepted D12633: match sendfile() error handling to send().

I'm sorry for accepting the previous patch too fast, without investing time into proper review.

Oct 10 2017, 8:03 PM

glebius accepted D12633: match sendfile() error handling to send().

My suggestion is to commit in two patches. First, revert r324405. Second, move of SS_ISCONNECTED check.

Oct 10 2017, 12:15 AM

Oct 9 2017

glebius committed rS324448: Improvements to sendfile(2) mbuf free routine..

Improvements to sendfile(2) mbuf free routine.

Oct 9 2017, 9:06 PM

glebius committed rS324447: In mb_dupcl() don't copy full m_ext, to avoid cache miss. Respectively,.

In mb_dupcl() don't copy full m_ext, to avoid cache miss. Respectively,

Oct 9 2017, 8:52 PM

glebius closed D12615: Mbuf external storage improvements..

Oct 9 2017, 8:35 PM

glebius committed rS324446: Shorten list of arguments to mbuf external storage freeing function..

Shorten list of arguments to mbuf external storage freeing function.

Oct 9 2017, 8:35 PM

glebius added a comment to D12615: Mbuf external storage improvements..

The change doesn't affect sfbufs at all.

Oct 9 2017, 2:29 PM

Oct 7 2017

glebius added a comment to D12615: Mbuf external storage improvements..

Yes, it will be. It is split in commits in my git repo.

Oct 7 2017, 3:29 PM

Oct 6 2017

glebius committed rS324374: Declare pmtud_blackhole global variables in tcp_timer.h, so that.

Declare pmtud_blackhole global variables in tcp_timer.h, so that

Oct 6 2017, 8:33 PM

glebius created D12615: Mbuf external storage improvements..

Oct 6 2017, 8:17 PM

glebius accepted D12575: Check so_error early in sendfile() call.

Oct 6 2017, 8:09 PM

glebius added a comment to D12551: ifnet(9): split ifc_alloc_unit() (should make code a bit cleaner)..

Can you please grab the PR? :)

Oct 6 2017, 8:07 PM

Oct 2 2017

glebius committed rP451103: Fix lsof build after src r324225..

Fix lsof build after src r324225.

Oct 2 2017, 11:31 PM

glebius committed rS324227: Hide struct socket and struct unpcb from the userland..

Hide struct socket and struct unpcb from the userland.

Oct 2 2017, 11:30 PM

glebius added inline comments to D12492: Ensure strict error handling of tcp_usr_send.

Oct 2 2017, 7:11 PM

glebius added a comment to D12501: count bytes in vn_sendfile when pru_send returns EAGAIN.

How is that possible that tcp_usr_send leads to tcp_connect and in_pcbbind? Shouldn't we get immediate ENOTCONN if we try to sendfile on a not connected socket?

Oct 2 2017, 6:41 PM

glebius added a comment to D12501: count bytes in vn_sendfile when pru_send returns EAGAIN.

That's an interesting finding! However, I don't like that we are blindly following write() or send() logic. Copy-pasting the ERESTART is definitely superfluous. I'm pretty sure that protocols shall never return that. More complex question is that why are we sure that EAGAIN confirms that all bytes sent were actually sent? Could be bug not in sendfile, but in write :) Let me dig more on that.

Oct 2 2017, 6:39 PM

glebius added a comment to D12551: ifnet(9): split ifc_alloc_unit() (should make code a bit cleaner)..

Does this fix this bug? https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=217401

Oct 2 2017, 6:27 PM