imp (Warner Losh)
User

Projects

User Details

User Since
Jun 2 2014, 4:20 PM (172 w, 4 d)

Recent Activity

Today

imp added a comment to D12466: Allow LOADER_TFTP_SUPPORT to work again.

lgtm

Fri, Sep 22, 3:43 PM
imp added a comment to D12465: Make MMCCAM work with sdhci_acpi.

Not sure I like another special case. Please consider this instead. It would also correctly detect this at compile time.

Fri, Sep 22, 3:39 PM
imp added a comment to D12434: Move MI bits of efirt to a MI file.

My eye didn't trip over anything...

Fri, Sep 22, 2:37 PM
imp accepted D12421: Compile loader as Little-Endian on PPC64/POWER8.

I like this. I'll echo the concerns of how to build both in one tree, but this allows either/or choice which is the next best thing.

Fri, Sep 22, 2:34 PM
imp committed rS323894: cam iosched: Bettar account IOPS for smoother performance.
cam iosched: Bettar account IOPS for smoother performance
Fri, Sep 22, 2:36 AM
imp committed rS323893: cam iosched: Enforce iop limits below the quanta value.
cam iosched: Enforce iop limits below the quanta value
Fri, Sep 22, 2:36 AM

Yesterday

imp committed rS323888: Add Chuck Tuffli (chuck@). He'll be working on CAM and NVMe related.
Add Chuck Tuffli (chuck@). He'll be working on CAM and NVMe related
Thu, Sep 21, 11:32 PM
imp committed rS323883: Always create usr/local/etc -> /etc/local symlink.
Always create usr/local/etc -> /etc/local symlink
Thu, Sep 21, 11:11 PM

Wed, Sep 20

imp committed rS323834: Fix queue depth for nda..
Fix queue depth for nda.
Wed, Sep 20, 9:42 PM
imp committed rS323832: cam iosched: Call cam_iosched_limiter_init() after ios->current is set to the….
cam iosched: Call cam_iosched_limiter_init() after ios->current is set to the…
Wed, Sep 20, 9:26 PM
imp closed D12349: cam iosched: Call cam_iosched_limiter_init() after ios->current is set to the default.
Wed, Sep 20, 9:26 PM
imp committed rS323831: cam iosched: Schedule cam_iosched_ticker() quanta times per second.
cam iosched: Schedule cam_iosched_ticker() quanta times per second
Wed, Sep 20, 9:26 PM
imp closed D12350: cam iosched: Schedule cam_iosched_ticker() quanta times per second.
Wed, Sep 20, 9:26 PM
imp accepted D12349: cam iosched: Call cam_iosched_limiter_init() after ios->current is set to the default.

lgtm

Wed, Sep 20, 9:21 PM
imp accepted D12350: cam iosched: Schedule cam_iosched_ticker() quanta times per second.

looks good.

Wed, Sep 20, 9:20 PM
imp committed rS323829: cam iosched: Add a handler for the quanta sysctl to enforce valid values.
cam iosched: Add a handler for the quanta sysctl to enforce valid values
Wed, Sep 20, 9:20 PM
imp closed D12351: cam iosched: Add a handler for the quanta sysctl to enforce valid values.
Wed, Sep 20, 9:20 PM
imp accepted D12351: cam iosched: Add a handler for the quanta sysctl to enforce valid values.

You listened to my prior feedback. This is perfect now. Thanks!

Wed, Sep 20, 9:14 PM
imp committed rS323828: cam iosched: Use the write queue for BIO_ZONE commands.
cam iosched: Use the write queue for BIO_ZONE commands
Wed, Sep 20, 9:13 PM
imp closed D12353: cam iosched: Use the write queue for BIO_ZONE commands.
Wed, Sep 20, 9:13 PM
imp added a comment to D12353: cam iosched: Use the write queue for BIO_ZONE commands.

On second thought, when I commit this, I'll change it to bp->bio_cmd != BIO_READ.

Wed, Sep 20, 9:07 PM
imp accepted D12353: cam iosched: Use the write queue for BIO_ZONE commands.

This works. However, it points to a larger issue with the cam iosched. We're not careful about honoring BIO_ORDERED commands....

Wed, Sep 20, 8:58 PM
imp added inline comments to D12422: loader: support for mixed-endianness ELF/loader and POWER8.
Wed, Sep 20, 1:23 PM
imp added inline comments to D12421: Compile loader as Little-Endian on PPC64/POWER8.
Wed, Sep 20, 1:14 PM

Mon, Sep 18

imp accepted D12386: Add kern.features flag for MMCCAM.
Mon, Sep 18, 2:45 PM
imp accepted D12111: Make basic Broadcom I/O space reading functions work.
Mon, Sep 18, 2:45 PM
imp accepted D12368: loader: biosmem allocate heap just below 4GB.
Mon, Sep 18, 1:18 PM

Fri, Sep 15

imp committed rS323625: Allow multiple TRIMs to be done for nda.
Allow multiple TRIMs to be done for nda
Fri, Sep 15, 8:16 PM
imp committed rS323624: Update comments on what the CAM_IOSCHED_FLAG_TRIM_ACTIVE means..
Update comments on what the CAM_IOSCHED_FLAG_TRIM_ACTIVE means.
Fri, Sep 15, 8:16 PM
imp added a comment to D12383: Fix problems with iflib when !EARLY_AP_STARTUP.

Fails for me:

Fri, Sep 15, 7:36 PM
imp created D12379: Pull in all non-srandom, non-pledge patches from OpenBSD.
Fri, Sep 15, 1:46 PM

Thu, Sep 14

imp accepted D12370: loader: biosmem.c cstyle cleanup.

my eyeballs didn't trip over anything here...

Thu, Sep 14, 3:23 PM
imp accepted D12368: loader: biosmem allocate heap just below 4GB.

Change looks OK, but I agree with Ed, it would be better to do the white space changes separately.

Thu, Sep 14, 1:59 PM
imp committed rS323577: Implement gawk multiple-arg extension to and, or, and xor..
Implement gawk multiple-arg extension to and, or, and xor.
Thu, Sep 14, 5:48 AM
imp committed rS323576: Bring in bit operation functions, ala gawk..
Bring in bit operation functions, ala gawk.
Thu, Sep 14, 5:48 AM
imp closed D12361: Bring in bit operation functions, ala gawk..
Thu, Sep 14, 5:48 AM
imp accepted D12364: Add PNP metadata to a few drivers.

Thanks for the changes. Now that my eyes are uncrossed, this looks sane to me now.

Thu, Sep 14, 5:30 AM
imp added inline comments to D12364: Add PNP metadata to a few drivers.
Thu, Sep 14, 4:56 AM
imp added a comment to D12361: Bring in bit operation functions, ala gawk..

OK. All the bugs I found are fixed, and the gawk extensions are now implemented more fully. I intend to commit this, unless I hear otherwise.

Thu, Sep 14, 4:47 AM
imp updated the diff for D12361: Bring in bit operation functions, ala gawk..

back to master...

Thu, Sep 14, 4:46 AM
imp updated the diff for D12361: Bring in bit operation functions, ala gawk..

Looks like someone s/0/NULL in the context of a pointer,
so make it look the same.

Thu, Sep 14, 4:45 AM
imp updated the diff for D12361: Bring in bit operation functions, ala gawk..

Implement the gawk aribtrary number args for and, or, and xor.

Thu, Sep 14, 4:42 AM
imp updated the diff for D12361: Bring in bit operation functions, ala gawk..

Minor tweaks: fix error messages and pendantic white space issues.

Thu, Sep 14, 3:35 AM
imp added a comment to D12361: Bring in bit operation functions, ala gawk..
In D12361#256383, @pfg wrote:

Minor tab/space issues but otherwise looks fine. Thanks!

Thu, Sep 14, 3:31 AM
imp added inline comments to D12364: Add PNP metadata to a few drivers.
Thu, Sep 14, 3:17 AM

Wed, Sep 13

imp created D12361: Bring in bit operation functions, ala gawk..
Wed, Sep 13, 8:16 PM
imp added inline comments to D12355: MFV r323535: 8585 improve batching done in zil_commit().
Wed, Sep 13, 7:44 PM
imp accepted D12357: New committer setup for chuck@.
Wed, Sep 13, 5:06 PM
imp committed rS323519: Minor fixes to edge cases in efi_get_next_variable_name.
Minor fixes to edge cases in efi_get_next_variable_name
Wed, Sep 13, 4:32 AM
imp added a comment to D10931: boot1 generate-fat: generate all templates at once.

I'm unsure as to what needs to happen now. Do I need to do anything to my patches yet?

We have D9860 open still which needs one more round of questions answered and updates, and then I'll commit that. After that I think it's all your patches.

Wed, Sep 13, 12:16 AM
imp added a comment to D10931: boot1 generate-fat: generate all templates at once.

I'm unsure as to what needs to happen now. Do I need to do anything to my patches yet?

We have D9860 open still which needs one more round of questions answered and updates, and then I'll commit that. After that I think it's all your patches.

Wed, Sep 13, 12:16 AM

Tue, Sep 12

imp committed rS323511: cam iosched: Limit the quanta default to hz if it's below 200.
cam iosched: Limit the quanta default to hz if it's below 200
Tue, Sep 12, 11:46 PM
imp closed D12337: cam iosched: Limit the quanta default to hz if it's below 200.
Tue, Sep 12, 11:46 PM
imp added a comment to D11927: Cavium LiquidIo 2350/2360 10/25G Intelligent NIC driver Initial submission.

Since this driver explicitly refuses to compile on non-64bit architectures (where LP64 is not defined), I'm moving this to files.amd64 as I suspect that this was what was intended. Else this breaks every FreeBSD build.

Tue, Sep 12, 11:43 PM
imp accepted D12337: cam iosched: Limit the quanta default to hz if it's below 200.

This looks good, but isn't the patch I expected from private email. I'll take a dry-run at committing it.

Tue, Sep 12, 10:48 PM
imp accepted D12317: Deorbit catman. It's useless on modern hardware..

Now all you need to do is go through the tree looking for all the places that cartman is an alias for catman...

Tue, Sep 12, 9:55 PM
imp committed rS323500: End softfp->hardfp transition period for arm.
End softfp->hardfp transition period for arm
Tue, Sep 12, 5:06 PM
imp accepted D12340: libefi: efipart_open should check the status from disk_open.
Tue, Sep 12, 2:16 PM
imp accepted D9463: gmirror: treat ENXIO as disk disconnect, not media error.

This looks good to my eye

Tue, Sep 12, 2:06 PM
imp added inline comments to D12274: End softfp->hardfp transition period for arm.
Tue, Sep 12, 2:03 PM

Mon, Sep 11

imp accepted D12111: Make basic Broadcom I/O space reading functions work.
Mon, Sep 11, 8:56 PM
imp accepted D12110: Add MMCCAM-enabled kernel config for IMX6, reduce debug noice in MMCCAM kernels.
Mon, Sep 11, 8:56 PM
imp accepted D12109: Miscellaneous fixes and improvements to MMCCAM stack.
Mon, Sep 11, 8:55 PM
imp added inline comments to D9680: Increase EFI MSDOSFS image size to 512Kib.
Mon, Sep 11, 4:08 PM
imp added a comment to D12315: Update bsdinstall/zfsboot to use newfs_msdos instead of an image file.

Apart from the fact that this enshrines too much knowledge of how to make things bootable, this is fine (since it already knows and my stuff isn't quite ready).

Mon, Sep 11, 4:06 AM

Sun, Sep 10

imp accepted D12304: Add information about new src committer (kibab).
Sun, Sep 10, 8:12 PM
imp accepted D12303: Introduce information about new commiter - kibab.

Looks good to my eye.

Sun, Sep 10, 8:12 PM
imp updated the diff for D12302: Move EFI FAT creation to here to obviate need for the boot1.efifat file.

Move all the white space changes to a separate commit
Fix some issues with returning the wrong thing
Simplify the system call (though I think we may just want to have a popen

to a script that knows all the cases and rip out most of the knowledge
from here so all creators of images / disks can use it).
Sun, Sep 10, 5:20 PM
imp added inline comments to D12302: Move EFI FAT creation to here to obviate need for the boot1.efifat file.
Sun, Sep 10, 2:50 PM
imp added a comment to D12302: Move EFI FAT creation to here to obviate need for the boot1.efifat file.

Let the comments begin :)

Sun, Sep 10, 5:41 AM
imp added reviewers for D12302: Move EFI FAT creation to here to obviate need for the boot1.efifat file: emaste, allanjude, nwhitehorn.
Sun, Sep 10, 5:36 AM
imp accepted D10931: boot1 generate-fat: generate all templates at once.

This looks good, so far as it goes... But maybe effort would be better fixing bsdinstall to not suck so bad as to need this as one file it DDs into a partition.

Sun, Sep 10, 5:34 AM
imp created D12302: Move EFI FAT creation to here to obviate need for the boot1.efifat file.
Sun, Sep 10, 5:33 AM

Sat, Sep 9

imp added a comment to D10447: Unify boot1 with loader.

! In D10447#254852, @eric_metricspace.net wrote:
That was because efi_main.c got moved out of libefi to loader. I moved it back.

Sat, Sep 9, 9:43 PM
imp added a comment to D10931: boot1 generate-fat: generate all templates at once.

Note: my question isn't a 'no' here unless the answer to the question is 'it totally works, what rock have you been under to not know that." :)

Sat, Sep 9, 9:36 PM
imp added a comment to D10931: boot1 generate-fat: generate all templates at once.

Once again I ask: where are we with makefs and FAT filesystems? :) Then we could kill these dreadful templates once and for all.

Sat, Sep 9, 9:35 PM
imp committed rS323379: It's been pointed out that init_script at least is useful w/o.
It's been pointed out that init_script at least is useful w/o
Sat, Sep 9, 9:34 PM
imp committed rS323375: Don't build uart_dev_mvebu unless we're on arm64..
Don't build uart_dev_mvebu unless we're on arm64.
Sat, Sep 9, 8:14 PM
imp added a comment to D12295: Kill gets().

Wouldn't it be better to just remove gets.c from Makefile.inc?

Sat, Sep 9, 7:10 PM
imp accepted D12287: loader.efi: chain loader should provide proper device handle.

This looks good to me.

Sat, Sep 9, 5:57 PM
imp committed rS323367: Mark init_chroot and init_script variables as deprecated..
Mark init_chroot and init_script variables as deprecated.
Sat, Sep 9, 4:05 PM
imp added a comment to D12010: Support armv7 builds for userland.

Damn... You caught all the places I was being lazy :) I'll go back and try to fix them. Thanks!

Sat, Sep 9, 3:53 PM

Fri, Sep 8

imp added a comment to D12274: End softfp->hardfp transition period for arm.

Can be this MFCed to stable? In other words, do we support in place upgrade from 10 (soft FP ABI) to 11 (hard FP ABI)?
From my point of view, this is OK for FBSD 12.

Fri, Sep 8, 3:31 PM
imp added reviewers for D12274: End softfp->hardfp transition period for arm: ian, andrew, cognet, manu.
Fri, Sep 8, 3:10 PM
imp created D12274: End softfp->hardfp transition period for arm.
Fri, Sep 8, 3:08 PM

Thu, Sep 7

imp committed rS323272: Be consistent and do return (1);.
Be consistent and do return (1);
Thu, Sep 7, 3:47 PM
imp committed rS323261: Fix armv6 build.
Fix armv6 build
Thu, Sep 7, 7:30 AM
imp committed rS323260: Create efi utility printing routines.
Create efi utility printing routines
Thu, Sep 7, 7:30 AM
imp committed rS323259: Implement efidp_size.
Implement efidp_size
Thu, Sep 7, 7:30 AM
imp committed rS323258: ucs2len.
ucs2len
Thu, Sep 7, 7:30 AM

Wed, Sep 6

imp added a comment to D12238: Implement random modulus with kern.randompid=1.

It would be better to use a less predictable method of getting the next PID if you want random PIDs...

Wed, Sep 6, 2:13 AM

Tue, Sep 5

imp committed rS323169: Add Ilya Babukin (kibab) to the src committers. He'll be looking after.
Add Ilya Babukin (kibab) to the src committers. He'll be looking after
Tue, Sep 5, 1:41 AM

Sun, Sep 3

imp added a comment to D12194: Fix loader bug causing too many pages allocation when bootloader is U-Boot.

I have no objection, but am curious if we ever need to honor the mmsz.

Sun, Sep 3, 3:13 PM · arm64
imp accepted D12219: libefivar: -fno-strict-aliasing.

Love it.

Sun, Sep 3, 3:10 PM

Sat, Sep 2

imp committed rS323131: Revert r322941: Eliminate redundant device matching functions.
Revert r322941: Eliminate redundant device matching functions
Sat, Sep 2, 6:18 PM
imp committed rS323125: The hard drive media device path contains the size of the partition,.
The hard drive media device path contains the size of the partition,
Sat, Sep 2, 7:04 AM
imp added a comment to D12211: libefivar: avoid type-punning in DevPathFromTextFibreEx.
In D12211#253281, @imp wrote:

This is upstream code.
The warning should be disabled instead.

Whoops, didn't realize. I see lib/libefivar/FreeBSD-update now. So, am I reading it right then to understand that efivar-dp-parse.c is from an upstream, but uefi-dplib.h (modified in D12210) is just in FreeBSD?

I'll check if there's anything else to be done with the flags when building this.

Sat, Sep 2, 12:22 AM
imp requested changes to D12211: libefivar: avoid type-punning in DevPathFromTextFibreEx.

This is upstream code.
The warning should be disabled instead.

Sat, Sep 2, 12:12 AM

Fri, Sep 1

imp committed rS323108: Add efimedia attribute for all GPT partitions..
Add efimedia attribute for all GPT partitions.
Fri, Sep 1, 5:55 PM
imp closed D12206: Add efimedia attribute for all GPT partitions. by committing rS323108: Add efimedia attribute for all GPT partitions..
Fri, Sep 1, 5:55 PM