imp (Warner Losh)
User

Projects

User Details

User Since
Jun 2 2014, 4:20 PM (181 w, 4 d)

Recent Activity

Yesterday

imp added a comment to D13224: add BIO_NORETRY flag, implement support in ata_da, use in ZFS vdev_geom.

After reading the links and ing.gila_gmail.com's description, it looks like FAILFAST and NORETRY are two different things. FAILFAST will faill all other I/O if one I/O fails. NORETRY will not retry the I/O when the drive tells us to retry the I/O. The ZFS code expects the former semantics, not the NORETRY semantics.

Fri, Nov 24, 5:55 PM
imp added a reviewer for D13224: add BIO_NORETRY flag, implement support in ata_da, use in ZFS vdev_geom: scottl.
Fri, Nov 24, 1:00 PM
imp requested changes to D13224: add BIO_NORETRY flag, implement support in ata_da, use in ZFS vdev_geom.

I think this is a terrible idea, at least as implemented. It suffers several problems.

Fri, Nov 24, 1:00 PM
imp committed rS326144: Mark the func pointer as __dead2. It looks up loader_main, which.
Mark the func pointer as __dead2. It looks up loader_main, which
Fri, Nov 24, 5:01 AM
imp committed rS326143: Fix theoretical integer overflow issues. If the product here is.
Fix theoretical integer overflow issues. If the product here is
Fri, Nov 24, 5:00 AM

Wed, Nov 22

imp added inline comments to D13198: Fix sign-extension error in adagetparams.
Wed, Nov 22, 10:39 PM
imp accepted D13194: Always null-terminate CAM periph_name and dev_name.

I didn't tag all the places that need it, but I'd be tempted to go with sizeof(dst) rather than a #define that may or may not be right and requires extra brain power to go check for each case.

Wed, Nov 22, 6:27 PM
imp accepted D13135: Fix fabs() for MIPS when used on -0.0.

This is honestly fine as it is. Two minor nits, but they are uber nits and can be ignored perhaps...

Wed, Nov 22, 4:31 PM
imp committed rS326085: Add a reference to getfsstat in statfs..
Add a reference to getfsstat in statfs.
Wed, Nov 22, 5:27 AM

Tue, Nov 21

imp committed rS326072: Unbreak riscv build in universe..
Unbreak riscv build in universe.
Tue, Nov 21, 7:23 PM
imp committed rS326071: Use TARGET_ARCH=riscv64 when TARGET=riscv.
Use TARGET_ARCH=riscv64 when TARGET=riscv
Tue, Nov 21, 7:23 PM
imp committed rS326069: Fix gptzfsboot for cases with GELI..
Fix gptzfsboot for cases with GELI.
Tue, Nov 21, 6:04 PM
imp committed rS326051: This program is more useful if it skips leading whitespace when.
This program is more useful if it skips leading whitespace when
Tue, Nov 21, 7:35 AM
imp committed rS326050: Document what the command line arguments actually do. List some of the.
Document what the command line arguments actually do. List some of the
Tue, Nov 21, 7:35 AM
imp committed rS326049: While the EFI spec allows numbers to be in many forms, libefivar.
While the EFI spec allows numbers to be in many forms, libefivar
Tue, Nov 21, 6:12 AM
imp committed rS326048: Remove trailing whitespace (one I just introduced and a bunch of.
Remove trailing whitespace (one I just introduced and a bunch of
Tue, Nov 21, 5:42 AM
imp committed rS326047: Implement efi media tagging for MBR partitioning types..
Implement efi media tagging for MBR partitioning types.
Tue, Nov 21, 5:35 AM

Mon, Nov 20

imp committed rS326038: Move some more common stuff up to Makefile.inc. In particular, the no.
Move some more common stuff up to Makefile.inc. In particular, the no
Mon, Nov 20, 10:42 PM
imp committed rS326037: Add -mno-avx2 for clang as well as -mno-avx. We don't want either of.
Add -mno-avx2 for clang as well as -mno-avx. We don't want either of
Mon, Nov 20, 10:41 PM
imp accepted D13106: da(4): Short-circuit unnecessary BIO_FLUSH commands.

As far as I can tell, this rev is fine. I get the ada issues and while I would like to see the same sort of thing done there, I accept that it's harder.

Mon, Nov 20, 10:21 PM
imp accepted D13173: Fix multiple bugs in cam_strmatch.

I suspect these same bugs are present in the file globbing code I stole the original implementation from.
Nice catches.

Mon, Nov 20, 9:46 PM
imp accepted D13171: print ZBC device type in "camcontrol identify" output.
Mon, Nov 20, 7:32 PM
imp added a comment to D13156: Remove hacks to support lint from system headers and MD x86 headers..

Looks good. One comment about LINT and one minor nit with varargs for super-duper old-school compilers we don't support anyway.

Mon, Nov 20, 4:25 PM

Fri, Nov 17

imp committed rS325954: Remove build system support for lint..
Remove build system support for lint.
Fri, Nov 17, 6:16 PM
imp closed D13124: Remove build system support for lint..
Fri, Nov 17, 6:16 PM
imp accepted D13133: Add SPDX license tags for 4 Clause BSD Licensed files..

I spot checked about a dozen of these. I see no issues.

Fri, Nov 17, 3:35 PM
imp added a comment to D13124: Remove build system support for lint..

Next task may be removing -DLINT support in source files like:

Fri, Nov 17, 5:15 AM
imp updated the diff for D13124: Remove build system support for lint..

bdrewery's fixes...

Fri, Nov 17, 4:43 AM
imp added a comment to D13125: Fix 'local' to not look in the source tree for the file..

this sounds reasonable to me...

Fri, Nov 17, 4:29 AM

Thu, Nov 16

imp created D13124: Remove build system support for lint..
Thu, Nov 16, 9:52 PM
imp committed rS325903: Only try to enable CK_CLYGRP if we're running on kernel newer than.
Only try to enable CK_CLYGRP if we're running on kernel newer than
Thu, Nov 16, 9:28 PM
imp closed D13114: Add sanity check for CK_CLYGRP.
Thu, Nov 16, 9:28 PM
imp accepted D12650: Add EFI RTC support.

this looks good to my eye... Stretch goal would be to allow adding clocks at runtime, but it's unclear what that would mean or how useful that would be given our current use cases....

Thu, Nov 16, 8:40 PM
imp updated the diff for D13114: Add sanity check for CK_CLYGRP.

Define IN_RTLD to pickup P_OSREL stuff

Thu, Nov 16, 7:38 PM
imp added a comment to D13114: Add sanity check for CK_CLYGRP.
In D13114#272815, @kib wrote:

How does this compile without -DIN_RTLD ?

Thu, Nov 16, 7:07 PM
imp added inline comments to D13114: Add sanity check for CK_CLYGRP.
Thu, Nov 16, 6:35 PM
imp updated the diff for D13114: Add sanity check for CK_CLYGRP.

Use correct version

Thu, Nov 16, 6:34 PM
imp added a comment to D12765: newfs: warn if newer than kernel.
In D12765#272641, @imp wrote:

Won't this trigger every single time we build a snapshot with a kernel that doesn't match?

This triggered because we ran new newfs, mounted that filesystem on an old kernel to populate it. Then the filesystem is used by the new kernel in the snapshot.

Thu, Nov 16, 4:10 PM
imp updated the summary of D13114: Add sanity check for CK_CLYGRP.
Thu, Nov 16, 3:53 PM
imp updated the diff for D13114: Add sanity check for CK_CLYGRP.

Fix per kib's comments.

Thu, Nov 16, 3:52 PM
imp added inline comments to D13114: Add sanity check for CK_CLYGRP.
Thu, Nov 16, 3:40 PM
imp added a comment to D12765: newfs: warn if newer than kernel.

Won't this trigger every single time we build a snapshot with a kernel that doesn't match? Unless we're rebooting with the new kernel we just built to build a snapshot, I'm having trouble understanding how this would be helpful in preventing the checksum issue from repeating when kirk, say, adds the same to the superblock or some other area he has bits defined for.

Thu, Nov 16, 5:58 AM
imp added a comment to D12765: newfs: warn if newer than kernel.

https://reviews.freebsd.org/D13114 is a better solution, imho.

Thu, Nov 16, 5:42 AM
imp added reviewers for D13114: Add sanity check for CK_CLYGRP: scottl, mckusick, emaste, kib, cem.
Thu, Nov 16, 5:35 AM
imp created D13114: Add sanity check for CK_CLYGRP.
Thu, Nov 16, 5:33 AM
imp committed rS325882: Use better wording: change there to the and define to defines..
Use better wording: change there to the and define to defines.
Thu, Nov 16, 12:20 AM

Wed, Nov 15

imp committed rS325881: Fix some formatting issues, bump .Dd to today's date, don't use.
Fix some formatting issues, bump .Dd to today's date, don't use
Wed, Nov 15, 11:51 PM
imp added a comment to D13106: da(4): Short-circuit unnecessary BIO_FLUSH commands.

Same thing for the ata version?

Wed, Nov 15, 5:00 PM
imp committed rS325855: Replace Fx's with 'the' since expanding FreeBSD here didn't seem quite.
Replace Fx's with 'the' since expanding FreeBSD here didn't seem quite
Wed, Nov 15, 3:02 PM
imp committed rS325854: Reword a bit for clarity..
Reword a bit for clarity.
Wed, Nov 15, 3:00 PM
imp accepted D13026: loader.efi: efipart does not recognize partitionless disks.

This looks OK, but the code could use some high level comments outlining the algorithms used since they are becoming harder to discern from the just reading the code.

Wed, Nov 15, 2:55 PM
imp added a comment to D13093: cam/scsi_da: active ssd quirks for all SAMSUNG MZ7* ssd.

Now that you point it, yes I missed that part!
It will be difficult for me to test it, do you want that I path ata_da.c anyway ?

Wed, Nov 15, 2:46 PM
imp committed rS325839: Fix SYSDIR path. After the move, we need to chop off a couple ../ from.
Fix SYSDIR path. After the move, we need to chop off a couple ../ from
Wed, Nov 15, 3:47 AM
imp committed rS325838: Inline pcie_link_{status,caps} where needed. Remove them as they.
Inline pcie_link_{status,caps} where needed. Remove them as they
Wed, Nov 15, 2:25 AM
imp added a comment to D13098: Make interrupt driven config hooks configurable and predictable..

This looks good to me...

Wed, Nov 15, 12:47 AM
imp committed rS325836: Add note about where to use MACHINE (just src/stand and src/sys and a.
Add note about where to use MACHINE (just src/stand and src/sys and a
Wed, Nov 15, 12:06 AM

Tue, Nov 14

imp committed rS325834: Move sys/boot to stand. Fix all references to new location.
Move sys/boot to stand. Fix all references to new location
Tue, Nov 14, 11:02 PM
imp committed rS325826: Move sys/boot/fdt/dts to sys/dts and adjust scripts..
Move sys/boot/fdt/dts to sys/dts and adjust scripts.
Tue, Nov 14, 9:04 PM
imp accepted D12762: Implement initial MMC partitions support for MMCCAM.

This looks OK-ish. There's still a fair amount of cut and paste going on, but that's no different than other drivers. There's a few things we can cleanup, but maybe in a separate commit.
After the updates for Marius' suggestions, we're close enough to style(9) to be ready to go.
I'm clicking Accept, but will feel better of Marius double checks my belief.

Tue, Nov 14, 8:08 PM · ARM
imp added a comment to D13093: cam/scsi_da: active ssd quirks for all SAMSUNG MZ7* ssd.

don't we also need this same quirk in ata_da.c?

Tue, Nov 14, 7:42 PM
imp committed rS325796: Properly decode NVMe state of the drive and print out the information.
Properly decode NVMe state of the drive and print out the information
Tue, Nov 14, 5:05 AM
imp committed rS325794: Provide link speed data in XPT_GET_TRAN_SETTINGS. Provide full version.
Provide link speed data in XPT_GET_TRAN_SETTINGS. Provide full version
Tue, Nov 14, 5:05 AM
imp committed rS325795: Belatedly add opt_nvme.h to fix building nvme.ko outside of a kernel.
Belatedly add opt_nvme.h to fix building nvme.ko outside of a kernel
Tue, Nov 14, 5:05 AM
imp committed rS325793: Provide pcie_link_status and pcie_link_cap convenience functions..
Provide pcie_link_status and pcie_link_cap convenience functions.
Tue, Nov 14, 5:05 AM
imp committed rS325792: Move zstd from contrib to sys/contrib so it can be used in the.
Move zstd from contrib to sys/contrib so it can be used in the
Tue, Nov 14, 5:03 AM
imp committed rS325785: Add two new tunables / sysctls to controll reboot after panic:.
Add two new tunables / sysctls to controll reboot after panic:
Tue, Nov 14, 12:29 AM
imp closed D13042: Add two new tunables / sysctls to controll reboot after panic.
Tue, Nov 14, 12:29 AM

Mon, Nov 13

imp added a comment to D13042: Add two new tunables / sysctls to controll reboot after panic.

OK. I'm going to push this in as-is, since I've tested it and an confident.
I'll also make it an enum in a future commit.
And do that via a sysctl API that lets one define ENUMs with fixed mapping like this, including tunable support.

Mon, Nov 13, 10:59 PM
imp committed rS325780: Don't add /boot/dt*s* but /boot/dt*b*. Stupid think-o. /boot/dtb was.
Don't add /boot/dt*s* but /boot/dt*b*. Stupid think-o. /boot/dtb was
Mon, Nov 13, 9:26 PM
imp committed rS325779: Add /boot/dts to the list of default modules. The minimal arm and mips.
Add /boot/dts to the list of default modules. The minimal arm and mips
Mon, Nov 13, 9:23 PM
imp committed rS325775: Add loader.conf to the list of files that are MD..
Add loader.conf to the list of files that are MD.
Mon, Nov 13, 8:39 PM
imp accepted D13047: rpi-firmware: Update to latest version.
Mon, Nov 13, 7:07 PM
imp added a comment to D13042: Add two new tunables / sysctls to controll reboot after panic.
In D13042#271831, @jhb wrote:

This is probably fine. You could also perhaps have a variant of kib@'s suggestion that takes a string: kern.panic_shutdown = {"reboot", "poweroff", "power-cycle"}. Internally you would store it as a simple int / enum, but use a SYSCTL_PROC to map strings to the internal values. This also means you can't set both RB_POWEROFF and RB_POWERCYCLE (which the current patch lets you do) for which the behavior isn't clear (which one has precedence, etc.?)

Mon, Nov 13, 7:06 PM
imp committed rS325765: Add notes about overlapping copies..
Add notes about overlapping copies.
Mon, Nov 13, 5:05 PM
imp committed rS325749: Document MACHINE, MACHINE_ARCH, CPUTYPE, TARGET and TARGET_ARCH.
Document MACHINE, MACHINE_ARCH, CPUTYPE, TARGET and TARGET_ARCH
Mon, Nov 13, 12:32 AM
imp closed D12981: Document MACHINE, MACHINE_ARCH, TARGET and TARGET_ARCH historical use and general philosphy..
Mon, Nov 13, 12:32 AM
imp committed rS325748: Use proper include file. While <boot/userboot/userboot.h> works, it.
Use proper include file. While <boot/userboot/userboot.h> works, it
Mon, Nov 13, 12:30 AM
imp added a comment to D13063: Fix setting RCA for MMC cards.

A few nits. It seems reasonable, but I'm not sure what RCA == 2 means.

Mon, Nov 13, 12:12 AM

Sun, Nov 12

imp committed rS325743: Make sure the proper loader.rc gets installed..
Make sure the proper loader.rc gets installed.
Sun, Nov 12, 5:11 PM

Sat, Nov 11

imp added a comment to D10407: Modify zstd so it can be built into the kernel as well.

A more complete grep shows the kernel will also use ${S}/../COPYRIGHT if it exists, but won't if it doesn't. And that's only for a copyright notice inside a comment for vers.c. Otherwise, all other relative path things in the kernel and module builds land inside the kernel.
There is one oddball. In libtekken which lives in sys/tekken/ there is a reference libc, but it's a library that doesn't look to be built at all and looks to be userland code.

Sat, Nov 11, 10:43 PM
imp added a comment to D10407: Modify zstd so it can be built into the kernel as well.
In D10407#271258, @cem wrote:

I am on board with the idea of moving contrib/zstd to sys/contrib/zstd. (I think the kernel build already reaches outside of sys/ for some headers, if not .c files, though.) Maybe bde would complain about the abomination.

Sat, Nov 11, 10:30 PM
imp accepted D13048: u-boot-rpi: Update to u-boot-master.

This looks good to my eye. Might want to mention EFI? Or not... Also, suggest getting Ian's views before committing, but don't let that delay too long if he's busy.

Sat, Nov 11, 7:23 PM
imp accepted D13049: u-boot-rpi2: Update to u-boot-master.

need we mention how to EFI?

Sat, Nov 11, 7:22 PM
imp accepted D13022: Add sysutils/u-boot-sopine.

Looks good to me. I have a few quibbles I may retouch after commit.

Sat, Nov 11, 5:03 PM
imp added inline comments to D13042: Add two new tunables / sysctls to controll reboot after panic.
Sat, Nov 11, 4:37 PM
imp added a comment to D13042: Add two new tunables / sysctls to controll reboot after panic.
In D13042#271028, @kib wrote:

May be, add a sysctl/tunable which takes the RB_XXX flags directly, instead of the per-flag variable, to be acted upon on panic ?

Sat, Nov 11, 4:35 PM
imp added a comment to D10407: Modify zstd so it can be built into the kernel as well.

It also occurs to me that moving this to the boot loader will mean we'll likely need to have the same sed'ing I suggested for the kernel, so there's opportunity for reuse of that. I'd be happy to help with that since that's clearly the next stop for these patches.

Sat, Nov 11, 4:31 PM
imp added inline comments to D10407: Modify zstd so it can be built into the kernel as well.
Sat, Nov 11, 4:28 PM
imp requested changes to D10407: Modify zstd so it can be built into the kernel as well.

I don't understand _BOOTSTRAP as an ifdef. We don't use that elsewhere. We use a couple of other things, but are trying to converge on _STAND
I don't understand why you aren't just creating a zstd-kernel.h file that has all the defined, including an appropriate malloc wrapper, and then sed'ing out the bad stuff and in the good stuff. libsa already does that for stand.h. See sys/boot/libsa/Makefile for details. Nothing in this diff seems to prelude that. You need to remove: <stdint.h> and replace <stddef.h>, <string.h>, <stdio.h> and <stdlib.h> I've suggested how to cope with malloc inline.

Sat, Nov 11, 4:25 PM
imp committed rS325718: Fix typo in filename..
Fix typo in filename.
Sat, Nov 11, 4:09 PM
imp added a comment to D12924: Remove mailaddr(7).

https://docs.freebsd.org/44doc/usd/07.mail/paper.pdf has The Mail Reference Manual.

Sat, Nov 11, 7:06 AM
imp added inline comments to D12924: Remove mailaddr(7).
Sat, Nov 11, 6:27 AM
imp accepted D12924: Remove mailaddr(7).

To: snapper!whipper!young!eadler
From: mylawn!getoff!imp
Subject: Modernization

Sat, Nov 11, 6:23 AM
imp updated the test plan for D13042: Add two new tunables / sysctls to controll reboot after panic.
Sat, Nov 11, 6:11 AM
imp updated the diff for D13042: Add two new tunables / sysctls to controll reboot after panic.

Update to include power cycle code... <doh>

Sat, Nov 11, 6:03 AM
imp created D13042: Add two new tunables / sysctls to controll reboot after panic.
Sat, Nov 11, 5:59 AM

Fri, Nov 10

imp committed rS325694: Install the 4th files in sys/boot/forth instead of each loader.
Install the 4th files in sys/boot/forth instead of each loader
Fri, Nov 10, 11:55 PM
imp committed rS325693: Move LOADER_{NO,}_GELI_SUPPORT to MK_LOADER_GELI.
Move LOADER_{NO,}_GELI_SUPPORT to MK_LOADER_GELI
Fri, Nov 10, 11:55 PM
imp committed rS325692: Replace LOADER_FIREWIRE_SUPPORT variable.
Replace LOADER_FIREWIRE_SUPPORT variable
Fri, Nov 10, 11:54 PM
imp committed rS325691: Remove useless PNP define here..
Remove useless PNP define here.
Fri, Nov 10, 11:54 PM