Page MenuHomeFreeBSD

jhibbits (Justin Hibbits)
User

Projects

User Details

User Since
May 16 2014, 7:29 PM (274 w, 2 d)

Recent Activity

Yesterday

jhibbits added a comment to D21301: powerpc: Link Book-E kernels at the same address as AIM kernels.

Fix 32-bit build.

Sun, Aug 18, 2:26 AM

Sat, Aug 17

jhibbits created D21301: powerpc: Link Book-E kernels at the same address as AIM kernels.
Sat, Aug 17, 3:44 AM
jhibbits added a comment to D21297: Fix loader on powerpc32.

It shouldn't be built for powerpc32, only powerpc64. However, it doesn't hurt anything, and is a good belt-and-suspenders change.

Sat, Aug 17, 2:13 AM · PowerPC

Fri, Aug 16

jhibbits added inline comments to D21286: Loader: Add load offset to powerpc kernel entry point.
Fri, Aug 16, 2:15 PM
jhibbits created D21286: Loader: Add load offset to powerpc kernel entry point.
Fri, Aug 16, 3:40 AM

Thu, Aug 15

jhibbits added a comment to D21206: libc: Add fdwalk(3), commonly used in Linux software.

Address @kib's feedback. The man page will probably need more iterating, but I
think the code is correct. It may not be optimal, though.

Thu, Aug 15, 3:44 AM
jhibbits committed rS351065: powerpc/pmap: Enable UMA_MD_SMALL_ALLOC for 64-bit booke.
powerpc/pmap: Enable UMA_MD_SMALL_ALLOC for 64-bit booke
Thu, Aug 15, 3:42 AM

Wed, Aug 14

jhibbits added a comment to D21206: libc: Add fdwalk(3), commonly used in Linux software.

Don't call getpid(), we can pass 0 now.

Wed, Aug 14, 1:50 AM
jhibbits added a comment to D21206: libc: Add fdwalk(3), commonly used in Linux software.

Incorporate mjg's changes.

Wed, Aug 14, 1:46 AM

Tue, Aug 13

jhibbits added a comment to D21206: libc: Add fdwalk(3), commonly used in Linux software.

Add a new sysctl, kern.proc.fdmap.<proc>, to provide the bitmap. Walk the bitmap instead of the file descriptor list.

Tue, Aug 13, 3:18 AM

Mon, Aug 12

D21233: Move sc out of the global file is now accepted and ready to land.
Mon, Aug 12, 9:02 PM
D21231: [ppc] avoid empty mdproc struct is now accepted and ready to land.
Mon, Aug 12, 5:10 PM
jhibbits committed rS350883: powerpc: Unify pmap definitions between AIM and Book-E.
powerpc: Unify pmap definitions between AIM and Book-E
Mon, Aug 12, 3:04 AM

Sun, Aug 11

jhibbits added a comment to D21206: libc: Add fdwalk(3), commonly used in Linux software.
In D21206#461101, @mjg wrote:

It's not hard to add such a sysctl. If you want I can hack it up for you. I think it should just export the fd bitmap, then iteration over it is trivial.

Sun, Aug 11, 3:41 AM

Sat, Aug 10

jhibbits added a comment to D21206: libc: Add fdwalk(3), commonly used in Linux software.

Address @kib's feedback. Man page will probably require several iterations
to get right. It's still a little muddy.

Sat, Aug 10, 8:10 PM
D21207: Move screen saver stuff to x86 NOTES is now accepted and ready to land.
Sat, Aug 10, 4:28 PM
D21208: Floppy driver really only works on x86 is now accepted and ready to land.
Sat, Aug 10, 4:26 PM
jhibbits created D21206: libc: Add fdwalk(3), commonly used in Linux software.
Sat, Aug 10, 2:53 AM

Fri, Aug 9

jhibbits committed rS350814: autofs: Fix autounmountd's printing of mount time..
autofs: Fix autounmountd's printing of mount time.
Fri, Aug 9, 2:20 AM

Thu, Aug 8

jhibbits added a comment to D15118: memcpy optimization for powerpc64.

Looks fine overall, just the one question. I just want to be sure this is thoroughly tested before it goes in, on the corner cases.

Thu, Aug 8, 9:46 PM · PowerPC
D20946: Bypass errno handling in _umtx_op_err on powerpc is now accepted and ready to land.

Looks good now.

Thu, Aug 8, 7:40 PM · PowerPC
jhibbits committed rS350756: mips: Fix register target for SP setting in libproc.
mips: Fix register target for SP setting in libproc
Thu, Aug 8, 4:37 PM
jhibbits closed D21188: Fix register target for SP setting in libproc.
Thu, Aug 8, 4:37 PM
jhibbits created D21188: Fix register target for SP setting in libproc.
Thu, Aug 8, 4:07 PM
jhibbits committed rS350738: powerpc/pmap: Minor optimizations to 64-bit booke pmap.
powerpc/pmap: Minor optimizations to 64-bit booke pmap
Thu, Aug 8, 3:18 AM
jhibbits committed rS350737: Change autounmountd(8) to use time_t for duration instead of double.
Change autounmountd(8) to use time_t for duration instead of double
Thu, Aug 8, 3:16 AM
jhibbits closed D21109: Change autounmountd(8) to use time_t for duration instead of double.
Thu, Aug 8, 3:16 AM

Tue, Aug 6

jhibbits committed rS350623: Fix build from r350622.
Fix build from r350622
Tue, Aug 6, 3:49 AM
jhibbits committed rS350622: powerpc/pmap: Simplify Book-E 64-bit page table management.
powerpc/pmap: Simplify Book-E 64-bit page table management
Tue, Aug 6, 3:16 AM

Mon, Aug 5

jhibbits committed rS350578: powerpc: Get 32-bit AIM building with secure-PLT.
powerpc: Get 32-bit AIM building with secure-PLT
Mon, Aug 5, 1:37 AM

Sun, Aug 4

jhibbits committed rS350573: Fix 32-bit build again, post r350570..
Fix 32-bit build again, post r350570.
Sun, Aug 4, 8:00 PM
jhibbits committed rS350572: Fix 32-bit build post-r350570.
Fix 32-bit build post-r350570
Sun, Aug 4, 7:55 PM
jhibbits committed rS350571: Fix the build after r350570..
Fix the build after r350570.
Sun, Aug 4, 7:47 PM
jhibbits closed D21096: Add necessary bits for Linux KPI to work correctly on powerpc.
Sun, Aug 4, 7:28 PM
jhibbits committed rS350570: Add necessary bits for Linux KPI to work correctly on powerpc.
Add necessary bits for Linux KPI to work correctly on powerpc
Sun, Aug 4, 7:28 PM
jhibbits committed rS350569: powerpc: Enable the BUS_DMA_NOCACHE flag handling.
powerpc: Enable the BUS_DMA_NOCACHE flag handling
Sun, Aug 4, 7:25 PM

Sat, Aug 3

D21136: [PPC64] Don't mark ld.bfd as obsolete is now accepted and ready to land.
Sat, Aug 3, 12:02 PM
D21131: [PPC64] Fix rtld direct exec mode is now accepted and ready to land.

Looks good. Can the same change apply to powerpc?

Sat, Aug 3, 11:58 AM
jhibbits committed rS350552: powerpc/powernv: Fix OPAL cfgread/cfgwrite error handling.
powerpc/powernv: Fix OPAL cfgread/cfgwrite error handling
Sat, Aug 3, 1:56 AM

Fri, Aug 2

jhibbits added a comment to D21096: Add necessary bits for Linux KPI to work correctly on powerpc.

Address @hselasky's comments.

Fri, Aug 2, 2:11 PM

Thu, Aug 1

jhibbits committed rS350494: powerpc/powernv: Only clear EEH freeze for some errors.
powerpc/powernv: Only clear EEH freeze for some errors
Thu, Aug 1, 3:59 AM
jhibbits committed rS350493: powerpc64/mmu: Use a SLIST for the PVO delete list, instead of a RB_TREE.
powerpc64/mmu: Use a SLIST for the PVO delete list, instead of a RB_TREE
Thu, Aug 1, 3:56 AM
jhibbits closed D21061: powerpc64/mmu: Use a SLIST for the PVO delete list, instead of a RB_TREE.
Thu, Aug 1, 3:56 AM

Wed, Jul 31

jhibbits added a comment to D20827: [PPC64] Implement CAS.

Thanks for checking. Looks good. I assume you tested on bare metal, too?

Actually, I've tested on VMs only, as this changes loader only, that is not used on POWER8 and above.
Do you want to test this on a G5? Or is there a way to test loader under PetitBoot?

Wed, Jul 31, 7:36 PM
D20827: [PPC64] Implement CAS is now accepted and ready to land.

Thanks for checking. Looks good. I assume you tested on bare metal, too?

Wed, Jul 31, 6:48 PM
jhibbits added a comment to D21096: Add necessary bits for Linux KPI to work correctly on powerpc.

@hselasky yes, I'm working on it, but got sidetracked doing other things while working on it. I'll have an update either tonight or tomorrow.

Wed, Jul 31, 2:20 PM

Tue, Jul 30

jhibbits added inline comments to D20827: [PPC64] Implement CAS.
Tue, Jul 30, 1:36 PM
jhibbits created D21109: Change autounmountd(8) to use time_t for duration instead of double.
Tue, Jul 30, 2:25 AM

Mon, Jul 29

jhibbits added inline comments to D21096: Add necessary bits for Linux KPI to work correctly on powerpc.
Mon, Jul 29, 9:34 PM

Sun, Jul 28

jhibbits added a comment to D21096: Add necessary bits for Linux KPI to work correctly on powerpc.

I'll have a closer look tomorrow. Most of this looks good.
Is this patch also needed for powerpc?
D21008

Sun, Jul 28, 9:44 PM
jhibbits created D21096: Add necessary bits for Linux KPI to work correctly on powerpc.
Sun, Jul 28, 9:03 PM

Thu, Jul 25

jhibbits created D21061: powerpc64/mmu: Use a SLIST for the PVO delete list, instead of a RB_TREE.
Thu, Jul 25, 3:48 AM
jhibbits committed rS350313: powerpc/pmap64: Make moea64 statistics optional.
powerpc/pmap64: Make moea64 statistics optional
Thu, Jul 25, 3:47 AM
jhibbits closed D20903: powerpc/pmap64: Make moea64 statistics optional.
Thu, Jul 25, 3:47 AM

Wed, Jul 24

jhibbits added inline comments to D15369: strncpy optimization for PowerPC64.
Wed, Jul 24, 3:39 PM · PowerPC
jhibbits added a comment to D15369: strncpy optimization for PowerPC64.

Have you checked performance on this latest version?

Wed, Jul 24, 2:50 PM · PowerPC
jhibbits added inline comments to D20903: powerpc/pmap64: Make moea64 statistics optional.
Wed, Jul 24, 2:46 PM
D21035: powerpc: Improve pvo allocation code is now accepted and ready to land.
Wed, Jul 24, 2:39 PM
jhibbits added a comment to D20903: powerpc/pmap64: Make moea64 statistics optional.

Make MOEA64_STATS a compile option

Wed, Jul 24, 3:07 AM
jhibbits committed rS350270: MFC r349874.
MFC r349874
Wed, Jul 24, 2:48 AM
jhibbits committed rS350269: MFC r346771.
MFC r346771
Wed, Jul 24, 2:37 AM

Tue, Jul 23

jhibbits added a reviewer for D21035: powerpc: Improve pvo allocation code: nwhitehorn.
Tue, Jul 23, 4:47 PM

Mon, Jul 22

jhibbits committed rS350232: powerpc: Unbreak 64-bit pmap from 350206.
powerpc: Unbreak 64-bit pmap from 350206
Mon, Jul 22, 10:59 PM
jhibbits committed rS350206: powerpc64/mmu: Make moea64_pvo_enter() return if an entry already exists.
powerpc64/mmu: Make moea64_pvo_enter() return if an entry already exists
Mon, Jul 22, 3:12 AM
jhibbits closed D20944: powerpc64/mmu: Make moea64_pvo_enter() return if an entry already exists.
Mon, Jul 22, 3:12 AM

Sun, Jul 21

jhibbits added a comment to D20944: powerpc64/mmu: Make moea64_pvo_enter() return if an entry already exists.

I just tested with a *very* small page table (0x8000 entries on a 256GB machine), and hit the 'moea64_pte_insert: overflow' panic, which you had mentioned a few years back is indicative of the PTEGs being filled with wired entries. Checking the statistics, I do see 0x38 overflows, and other statistics look accurate, too. I don't see anything bad in this patch, because the only thing it does is optimistically try to insert the PVO, and go through the normal logic if it already exists.

Sun, Jul 21, 4:01 AM
jhibbits committed rS350190: powerpc: Remove an unnecessary #ifdef guard from slb.c.
powerpc: Remove an unnecessary #ifdef guard from slb.c
Sun, Jul 21, 3:20 AM

Sat, Jul 20

jhibbits committed rS350181: powerpc/SPE: Enable SPV bit for EFSCFD instruction emulation.
powerpc/SPE: Enable SPV bit for EFSCFD instruction emulation
Sat, Jul 20, 6:22 PM

Jul 20 2019

jhibbits added a comment to D20944: powerpc64/mmu: Make moea64_pvo_enter() return if an entry already exists.

Just make the page table size really small. There's a spill counter in the statistics; when it gets small enough, you should see it moving. If spill-handling doesn't work, and you have spills, everything breaks fast.

Jul 20 2019, 3:22 AM

Jul 16 2019

jhibbits committed rS350023: powerpc: Fix casueword(9) post-r349951.
powerpc: Fix casueword(9) post-r349951
Jul 16 2019, 3:55 AM

Jul 15 2019

jhibbits added a comment to D20944: powerpc64/mmu: Make moea64_pvo_enter() return if an entry already exists.

Does this break PTE spills? This code path historically was involved in re-adding PVOs to the hashed page table on faults if they got spilled.

Jul 15 2019, 9:13 PM

Jul 14 2019

jhibbits added inline comments to D20946: Bypass errno handling in _umtx_op_err on powerpc.
Jul 14 2019, 3:53 AM · PowerPC

Jul 13 2019

jhibbits committed rS349965: powerpc64/pmap: No need for moea64_pvo_remove_from_page_locked() wrapper.
powerpc64/pmap: No need for moea64_pvo_remove_from_page_locked() wrapper
Jul 13 2019, 3:39 AM
jhibbits created D20944: powerpc64/mmu: Make moea64_pvo_enter() return if an entry already exists.
Jul 13 2019, 3:16 AM
jhibbits committed rS349963: powerpc64/pmap: Reduce scope of PV_LOCK in remove path.
powerpc64/pmap: Reduce scope of PV_LOCK in remove path
Jul 13 2019, 3:02 AM
jhibbits closed D20694: powerpc64/pmap: Reduce scope of PV_LOCK in remove path.
Jul 13 2019, 3:02 AM
jhibbits closed D20911: Set pcpu curpmap for powerpc64.
Jul 13 2019, 12:20 AM
jhibbits committed rS349960: Set pcpu curpmap for powerpc64.
Set pcpu curpmap for powerpc64
Jul 13 2019, 12:20 AM

Jul 12 2019

jhibbits closed D20811: Allow efi loader to get network params from uboot.
Jul 12 2019, 12:54 AM
jhibbits committed rS349928: Allow efi loader to get network params from uboot.
Allow efi loader to get network params from uboot
Jul 12 2019, 12:54 AM
jhibbits updated the summary of D20911: Set pcpu curpmap for powerpc64.
Jul 12 2019, 12:40 AM
jhibbits added a comment to D20911: Set pcpu curpmap for powerpc64.

It helps if I use an actual pointer.

Jul 12 2019, 12:39 AM
jhibbits added a comment to D20911: Set pcpu curpmap for powerpc64.

Simplify the code. Revert to the first version tested.

Jul 12 2019, 12:37 AM

Jul 11 2019

jhibbits added a comment to D20903: powerpc/pmap64: Make moea64 statistics optional.

I still see merit in keeping statistics available, for assessing performance issues as you mentioned, but I don't see it necessary in the general case. If there are no objections, I'll take @luporl's request and add an option for MOEA64_STATS, possibly enabled in CURRENT, but disabled in STABLE.

Jul 11 2019, 5:01 PM
jhibbits added a comment to D20903: powerpc/pmap64: Make moea64 statistics optional.

What do the statistics give us in general for pmap? I don't see statistics on other archs, so I'm not sure what the benefit is. I understood for bringup and early stabilization, but we're in pretty solid territory now.

Jul 11 2019, 4:23 PM
jhibbits added inline comments to D20827: [PPC64] Implement CAS.
Jul 11 2019, 4:22 PM
jhibbits created D20911: Set pcpu curpmap for powerpc64.
Jul 11 2019, 3:45 AM
jhibbits closed D20896: powerpc: Only worry about the lower 32 bits of SP in a 32-bit process.
Jul 11 2019, 3:29 AM
jhibbits committed rS349906: powerpc: Only worry about the lower 32 bits of SP in a 32-bit process.
powerpc: Only worry about the lower 32 bits of SP in a 32-bit process
Jul 11 2019, 3:29 AM

Jul 10 2019

jhibbits committed rS349874: powerpc: Clamp 32-bit binaries to 32-bit MAXUSER.
powerpc: Clamp 32-bit binaries to 32-bit MAXUSER
Jul 10 2019, 4:09 AM
jhibbits created D20903: powerpc/pmap64: Make moea64 statistics optional.
Jul 10 2019, 12:00 AM

Jul 9 2019

jhibbits created D20896: powerpc: Only worry about the lower 32 bits of SP in a 32-bit process.
Jul 9 2019, 10:00 PM
jhibbits added inline comments to D20827: [PPC64] Implement CAS.
Jul 9 2019, 9:33 PM
jhibbits added a comment to D20598: powerpc: Transition to Secure-PLT, like most other OSs.

Justin wrote "Secure-PLT is fully compatible with BSS-PLT, so there's no issue with compatibility".
Well, modern ld from devel/powerpc64-binutils exits with an error code when it reports things
like:
QUOTE
bss-plt forced due to /usr/obj/powerpcvtsc_clang_altbinutils/powerpc.powerpc/usr/src/powerpc.powerpc/tmp/usr/lib/crtbeginS.o
END QUOTE
The error code stops buildworld. (The old system binutils' ld does not exit with an error code and so does not stop the build.)

Jul 9 2019, 4:37 PM

Jul 8 2019

jhibbits added a comment to D20337: [PowerPC64] backport of LLVM fixes in preparation for building FreeBSD/PowerPC64.

This can now be abandoned, 8.0.1-rc3 was pulled in just the other day.

Jul 8 2019, 6:23 PM
D20857: [PPC64] pseries: fix realmaxaddr calculation is now accepted and ready to land.
Jul 8 2019, 4:24 PM

Jul 7 2019

jhibbits added a comment to D20857: [PPC64] pseries: fix realmaxaddr calculation.

Looks good overall. Just a couple nits..

Jul 7 2019, 6:08 PM

Jul 4 2019

jhibbits committed rS349742: MFC r347986:.
MFC r347986:
Jul 4 2019, 7:02 PM

Jul 3 2019

jhibbits committed rS349712: MFC r348299.
MFC r348299
Jul 3 2019, 9:30 PM
D20843: [PPC64] pseries llan: fix MAC address is now accepted and ready to land.
Jul 3 2019, 9:26 PM