imp (Warner Losh)
User

Projects

User Details

User Since
Jun 2 2014, 4:20 PM (207 w, 4 d)

Recent Activity

Today

imp added inline comments to D15481: Create a default wlan for wireless NICs if no wlans are specified..
Sat, May 26, 12:08 AM

Yesterday

imp accepted D15569: Fix building GENERIC-MMCCAM on arm64.
Fri, May 25, 3:26 PM · arm64
imp added inline comments to D15566: Make opal_pci driver work with POWER9.
Fri, May 25, 4:04 AM
imp accepted D15552: add support for console resuming, implement it for uart, use on x86.

This looks good.

Fri, May 25, 2:21 AM

Thu, May 24

imp committed rS334192: Protect bzero call against macro expansion.
Protect bzero call against macro expansion
Thu, May 24, 11:20 PM
imp committed rS334184: This is no unreferenced, so retire it..
This is no unreferenced, so retire it.
Thu, May 24, 9:11 PM
imp committed rS334183: Make memmove and bcopy share code.
Make memmove and bcopy share code
Thu, May 24, 9:11 PM
imp committed rS334182: Define memmove and make bcopy alt entry point.
Define memmove and make bcopy alt entry point
Thu, May 24, 9:11 PM
imp committed rS334181: Make memmove an alias for memcpy.
Make memmove an alias for memcpy
Thu, May 24, 9:11 PM
imp closed D15374: Various rearrangements to make memmove a first class citizen.
Thu, May 24, 9:11 PM
imp added a comment to D15374: Various rearrangements to make memmove a first class citizen.

I've considered Marius' suggestion. I reject it. memmove is now a first class citizen. If you'd like to make it even better, you are welcome to do so after this commit.

Thu, May 24, 9:11 PM
imp committed rS334166: We can't release the refcount outside of the periph lock..
We can't release the refcount outside of the periph lock.
Thu, May 24, 4:31 PM
imp closed D15517: We can't release the refcount outside of the periph lock..
Thu, May 24, 4:31 PM

Wed, May 23

imp added a comment to D15526: reduce overhead of entropy collection.

We only need to generate about 8 bytes of entropy per second total for the spiking of the algorithms...

Wed, May 23, 1:38 AM

Tue, May 22

imp added a comment to D15517: We can't release the refcount outside of the periph lock..
In D15517#327757, @mav wrote:

Looks good to me. The only rough edge I see is that outstanding_cmds is used only for the newly added assertion, that is why it was so broken before this and nobody cared.

Tue, May 22, 8:45 PM
imp abandoned D15453: Make reference counting more robust.

OK. So it looked easy...

Tue, May 22, 5:41 PM
imp added a reviewer for D15517: We can't release the refcount outside of the periph lock.: mav.
Tue, May 22, 5:38 PM
imp updated the diff for D15517: We can't release the refcount outside of the periph lock..

doh! no, with invariants

Tue, May 22, 5:36 PM
imp updated the diff for D15517: We can't release the refcount outside of the periph lock..

doh! compile with invariants

Tue, May 22, 5:34 PM
imp accepted D12762: Implement initial MMC partitions support for MMCCAM.
Tue, May 22, 4:52 PM · ARM
imp accepted D15521: Fix MMCCAM scanning for new cards.
Tue, May 22, 3:44 PM · ARM
imp added a comment to D15521: Fix MMCCAM scanning for new cards.

I agree. Comment says this was from nvme_scan_lun, and it uses path here not periph->path.

Tue, May 22, 3:43 PM · ARM

Mon, May 21

imp updated the diff for D15517: We can't release the refcount outside of the periph lock..

Regroup

Mon, May 21, 11:37 PM
imp updated the diff for D15517: We can't release the refcount outside of the periph lock..

fix bogus mdastart -> ndastart

Mon, May 21, 11:11 PM
imp updated the diff for D15517: We can't release the refcount outside of the periph lock..

Fix missed part of the patch

Mon, May 21, 11:09 PM
imp created D15517: We can't release the refcount outside of the periph lock..
Mon, May 21, 11:06 PM
imp added inline comments to D15514: Clarify in unlink(1) that like rm, unlink too requires "--".
Mon, May 21, 10:18 PM

Sun, May 20

imp committed rD51669: Fix preferred license..
Fix preferred license.
Sun, May 20, 3:01 AM

Sat, May 19

imp committed rS333880: Restore the all rights reserved language. Put it on each of the prior.
Restore the all rights reserved language. Put it on each of the prior
Sat, May 19, 5:30 PM

Fri, May 18

imp added a comment to D15483: More bcmp "optimization".

I have a straw-man patch that does some of what mjg is talking about. It doesn't remove the b* functions though. It's not quite ready for sharing, though if there's interest i'll polish up enough to publish.

Fri, May 18, 7:56 PM
imp accepted D15476: Quiesce build errors with cast to u_int8_t.
Fri, May 18, 5:15 PM
imp added inline comments to D15476: Quiesce build errors with cast to u_int8_t.
Fri, May 18, 5:05 PM
imp committed rS333798: Based on multiple sources including the State Gazette the last day of.
Based on multiple sources including the State Gazette the last day of
Fri, May 18, 4:20 PM
imp added inline comments to D15473: Add on bhyverun a new option '-B' that gives the capability to change the smbios table type1 using an ucl based configuration file..
Fri, May 18, 4:10 PM

Thu, May 17

imp accepted D15467: Use strsep() to parse init_path in start_init()..

OK. Can't find anything else to argue with :)

Thu, May 17, 10:49 PM
imp accepted D15467: Use strsep() to parse init_path in start_init()..

Despite a couple grumbles, this looks good to my eye.

Thu, May 17, 9:45 PM

Wed, May 16

imp committed rS333688: 2018 this time..
2018 this time.
Wed, May 16, 9:07 PM
imp added a reviewer for D15456: change ffs_blkfree() trim to not bypass geom_vfs, avoids panics: mav.
Wed, May 16, 7:33 PM
imp accepted D15456: change ffs_blkfree() trim to not bypass geom_vfs, avoids panics.

We can change to M_TRIM when Kirk commits his stuff :)

Wed, May 16, 7:32 PM
imp updated the test plan for D15453: Make reference counting more robust.
Wed, May 16, 7:30 PM
imp created D15453: Make reference counting more robust.
Wed, May 16, 3:50 PM
imp committed rS333675: Add note about LD=ld.lld being a temporary requirement when building.
Add note about LD=ld.lld being a temporary requirement when building
Wed, May 16, 1:52 PM
imp committed rS333673: MFC r333436: only launch getty if underlying device exists.
MFC r333436: only launch getty if underlying device exists
Wed, May 16, 1:18 PM

Tue, May 15

imp committed rS333651: Hold the reference count until the CCB is released.
Hold the reference count until the CCB is released
Tue, May 15, 10:22 PM
imp closed D15435: expand scope of da refcount to protect I/Os in flight.
Tue, May 15, 9:25 PM
imp committed rS333648: Hold the reference count until the CCB is released.
Hold the reference count until the CCB is released
Tue, May 15, 9:25 PM
imp added a comment to D15430: make unix socket locking finer grained.

I want to point out that the lack of code formatting tool like clang-format resulted in @markj spending some of his precious cycles reviewing not the substance of the patch, but instead commenting on style nits (correct according to style(9) - thank you @markj).

Using brain power for code formatting is the modern-day equivalent to creating fire by rubbing two sticks together instead of using a match/lighter. My comment isn't directed at anyone in particular, but is an observation from watching and participating in reviews in FreeBSD (vs other repos and orgs in my daily patch review queue where the review process is nearly 100% focused on the substance of the patch and not various nits that are caught by linting tools and formatters). I apologize if this sounds flippant, that is not my intent. I'm excited to see changes like these land but wish the friction of style(9) were obviated through modern tooling. This isn't a new comment to most on this review but I wanted to vent a little feedback.

Tue, May 15, 5:12 PM
imp accepted D15442: Retire vxge(4).

This approval is for a 'looks like you got it all and did the removal properly' aspect of this review.
I personally agree we should have removed this a while ago.

Tue, May 15, 4:29 PM

Mon, May 14

imp accepted D15435: expand scope of da refcount to protect I/Os in flight.

Looks good to me. One tiny question that occurred to me after I looked at it this morning. I think we're fine delaying the close for up to a tick so we don't have to try to send a bunch of wake ups during normal operation. Also did the obligatory lock check, and we're good there as well: refcount is only changed while holding the periph lock, and cam_periph_sleep uses the same lock to sleep on.

Mon, May 14, 8:34 PM
imp accepted D15415: [New Port] sysutils/dwatch-pwsnoop.

This looks good to my eye. I'd give mat@ 24 hours to opine on your changes in response to his comments.

Mon, May 14, 12:24 AM

Sun, May 13

imp added inline comments to D15415: [New Port] sysutils/dwatch-pwsnoop.
Sun, May 13, 5:17 PM
imp accepted D15407: [New Port] sysutils/dwatch-gource.

Looks good to me. I don't understand mat@'s comment. It looks stale, but if not my approval is contingent on you resolving it.

Sun, May 13, 4:20 PM

Sat, May 12

imp accepted D15401: sysutils/rpi-firware Update to 1.20180328.
Sat, May 12, 7:27 PM
imp committed rS333510: Remove extra copy of bcopy.c now that we're using the libkern version.
Remove extra copy of bcopy.c now that we're using the libkern version
Sat, May 12, 1:43 AM

Fri, May 11

imp added a comment to D15169: For video consoles, only launch a getty if the device exists..

Perhaps have a common file (can't help but notice it seems to be the only per-arch etc/ file, so we could get rid of the whole mechanism this way), and have per-image overrides specified by release/*/WHATEVER.conf files, like the rest of per-image things?

Fri, May 11, 10:15 PM
imp added a comment to D15387: Migrate top(1) to usr.bin.

looks like you've removed a lot of keywords...

Fri, May 11, 4:55 AM

Thu, May 10

imp accepted D14694: Silence down /etc/rc.d/kldxref.

This looks good. One minor concern about precedence, but if it's good, I'm good.

Thu, May 10, 8:50 PM
imp committed rS333471: Revert r333365.
Revert r333365
Thu, May 10, 8:27 PM
imp added a comment to D15378: Error on attempt to link amd64 with binutils 2.17.50 linker.
In D15378#324222, @kib wrote:

IMO it is more useful to check the ${LD} features, or at least the ${LD} version.

Thu, May 10, 3:28 PM
imp added a reviewer for D15374: Various rearrangements to make memmove a first class citizen: marius.
Thu, May 10, 3:04 PM
imp accepted D15365: simple preempt safe epoch API.
Thu, May 10, 1:34 PM
imp updated the diff for D15374: Various rearrangements to make memmove a first class citizen.

Update after committing the low-risk deck-chair motion.

Thu, May 10, 2:35 AM
imp reopened D15374: Various rearrangements to make memmove a first class citizen.

not quite done.

Thu, May 10, 2:35 AM
imp committed rS333445: Remove unused bcopyb..
Remove unused bcopyb.
Thu, May 10, 2:32 AM
imp committed rS333444: Simplify things a little.
Simplify things a little
Thu, May 10, 2:32 AM
imp committed rS333443: Move MI-ish bcopy routine to libkern.
Move MI-ish bcopy routine to libkern
Thu, May 10, 2:31 AM
imp closed D15374: Various rearrangements to make memmove a first class citizen.
Thu, May 10, 2:31 AM
imp added a comment to D15169: For video consoles, only launch a getty if the device exists..

I'm pretty sure we have MIPS boards (well, BERI running in FPGA, but still) with Terasic... screen-thingys with console on them, here at Cambridge. Not sure how it's configured regarding ttys(5), though, or if it even goes through vt(4) - although it certainly looks like a text console.

Thu, May 10, 12:54 AM
imp updated the diff for D15374: Various rearrangements to make memmove a first class citizen.

Update to actually remove libkern/mmemmove.c

Thu, May 10, 12:42 AM
imp created D15374: Various rearrangements to make memmove a first class citizen.
Thu, May 10, 12:38 AM

Wed, May 9

imp committed rS333436: For video consoles, only launch a getty if the device exists..
For video consoles, only launch a getty if the device exists.
Wed, May 9, 8:49 PM
imp closed D15169: For video consoles, only launch a getty if the device exists..
Wed, May 9, 8:49 PM
imp committed rS333435: Remove 'All Rights Reserved'.
Remove 'All Rights Reserved'
Wed, May 9, 8:41 PM
imp committed rS333434: Remove the 'All Rights Reserved' clause from some of the stuff I've.
Remove the 'All Rights Reserved' clause from some of the stuff I've
Wed, May 9, 8:32 PM
imp committed rS333433: Use the full year, for real this time..
Use the full year, for real this time.
Wed, May 9, 8:26 PM
imp updated the diff for D15371: Make camcontrol identify work with nda devices.

Mroe stuff?

Wed, May 9, 8:22 PM
imp created D15371: Make camcontrol identify work with nda devices.
Wed, May 9, 7:56 PM
imp committed rS333424: nda protocol rate reporting.
nda protocol rate reporting
Wed, May 9, 6:41 PM
imp accepted D15370: Fix copyright, remove "all rights reserved".
Wed, May 9, 4:25 PM
imp committed rS333409: Minor style nits.
Minor style nits
Wed, May 9, 2:11 PM
imp accepted D14694: Silence down /etc/rc.d/kldxref.

Almost right...

Wed, May 9, 1:53 PM
imp committed rD51636: Fix typo in Buenos Aires..
Fix typo in Buenos Aires.
Wed, May 9, 1:49 PM
imp added a comment to D15365: simple preempt safe epoch API.

Generally I like it.
I'd like it a lot more if there were man.9 page(s) for the epoch API.
Some nits are all I see, but my CK foo is weak.

Wed, May 9, 1:43 PM
imp committed rD51635: Add a note about the phrase 'All Rights Reserved' here to act as a.
Add a note about the phrase 'All Rights Reserved' here to act as a
Wed, May 9, 3:35 AM
imp added a comment to D14958: Add SPDX tags to example manual pages in share/examples/mdoc/.

You'll need to rebase after my push. Sorry.

Wed, May 9, 3:32 AM · manpages, Doc Committers
imp committed rS333391: Remove 'All Rights Reserved' from the collection copyright and templates..
Remove 'All Rights Reserved' from the collection copyright and templates.
Wed, May 9, 2:03 AM

Tue, May 8

imp committed rS333385: Remove ignored command line options.
Remove ignored command line options
Tue, May 8, 8:03 PM
imp committed rS333384: efibootmgr.8: fix example.
efibootmgr.8: fix example
Tue, May 8, 8:03 PM
imp accepted D15357: Recognize the base/gcc compiler as GCC..
Tue, May 8, 7:46 PM
imp committed rS333383: Improve printing the boot variables..
Improve printing the boot variables.
Tue, May 8, 7:44 PM
imp committed rS333381: Inline print_order(). It's used one palce..
Inline print_order(). It's used one palce.
Tue, May 8, 6:25 PM
imp committed rS333365: We don't use f_devdata, so don't set it. Should that need to change.
We don't use f_devdata, so don't set it. Should that need to change
Tue, May 8, 4:17 PM

Mon, May 7

imp added a comment to D14553: Set geom_eli module version.

I think you can close this.

Mon, May 7, 10:59 PM
imp committed rS333335: Put the CPU starting on one line..
Put the CPU starting on one line.
Mon, May 7, 9:09 PM
imp committed rS333334: Use device_quiet_children to silence verbose CPU probe messages..
Use device_quiet_children to silence verbose CPU probe messages.
Mon, May 7, 9:09 PM
imp committed rS333333: Add device_quiet_children() and device_has_quiet_children().
Add device_quiet_children() and device_has_quiet_children()
Mon, May 7, 9:09 PM
imp added a comment to D15326: Stop using non ctype.h versions of toupper().
In D15326#323223, @imp wrote:
In D15326#322966, @imp wrote:

So why isn't sys/ctypes.h enough? you're introducing a dependency on userland inside files that look like they are used by the kernel.

Ugh. This didn't occur to me as this was part of the buildworld stage. How about I add a "#ifndef _KERNEL #includ <ctype.h>" type thing. sys/ctype.h doesn't have all of the same ctype macros that are in /usr/include/ctype.h ... so that might be a problem.

why doesn't sys/ctype.h work? We should fix that problem...

Unless I'm super confused here, this is part of buildworld ... so _KERNEL isn't defined, therefore the include does nothing?

Mon, May 7, 5:40 PM
imp added a comment to D15326: Stop using non ctype.h versions of toupper().
In D15326#322966, @imp wrote:

So why isn't sys/ctypes.h enough? you're introducing a dependency on userland inside files that look like they are used by the kernel.

Ugh. This didn't occur to me as this was part of the buildworld stage. How about I add a "#ifndef _KERNEL #includ <ctype.h>" type thing. sys/ctype.h doesn't have all of the same ctype macros that are in /usr/include/ctype.h ... so that might be a problem.

Mon, May 7, 4:54 PM

Sun, May 6

imp added inline comments to D14553: Set geom_eli module version.
Sun, May 6, 5:58 PM
imp added a comment to D14553: Set geom_eli module version.

So there's already this at the end of the file:

Sun, May 6, 5:57 PM