Page MenuHomeFreeBSD

markj (Mark Johnston)
User

Projects (7)

User Details

User Since
Mar 12 2014, 1:00 AM (402 w, 6 d)

Recent Activity

Today

markj committed R10:5b5bbf2e7c34: sctp: Remove now-unneeded mb_unmapped_to_ext() calls (authored by markj).
sctp: Remove now-unneeded mb_unmapped_to_ext() calls
Tue, Nov 30, 1:37 AM
markj committed R10:422456ae27f5: sctp: Use m_apply() to calcuate a checksum for an mbuf chain (authored by markj).
sctp: Use m_apply() to calcuate a checksum for an mbuf chain
Tue, Nov 30, 1:36 AM
markj committed R10:02e3eb8d482d: mbuf: Only allow extpg mbufs if the system has a direct map (authored by markj).
mbuf: Only allow extpg mbufs if the system has a direct map
Tue, Nov 30, 1:36 AM

Yesterday

markj accepted D32657: bhyve: Fix the wrong Number of Power States Support value also set a reasonable Max Power for NVMe controller.

Sorry to nitpick a bit more. Please let me know if you'd like me to commit the change.

Mon, Nov 29, 9:48 PM
markj requested review of D33169: vmm: Permit writes to hw.vmm.destroy in capability mode.
Mon, Nov 29, 9:19 PM
markj committed R10:e0189cf66a94: elf.5: Document compressed section headers (authored by markj).
elf.5: Document compressed section headers
Mon, Nov 29, 6:58 PM
markj committed R10:9ff084f0fab6: libelf: Link gelf_getchdr.3 to the build (authored by markj).
libelf: Link gelf_getchdr.3 to the build
Mon, Nov 29, 6:58 PM
markj committed R10:7a3372303802: mii/dp83822phy: Remove a declaration for a non-existent function (authored by markj).
mii/dp83822phy: Remove a declaration for a non-existent function
Mon, Nov 29, 6:58 PM
markj committed R10:d5ea04ee7ba6: dummynet: Avoid an out-of-bounds read in do_config() (authored by markj).
dummynet: Avoid an out-of-bounds read in do_config()
Mon, Nov 29, 6:58 PM
markj committed R10:1c732c85911e: dummynet: Fix socket option length validation for IP_DUMMYNET3 (authored by markj).
dummynet: Fix socket option length validation for IP_DUMMYNET3
Mon, Nov 29, 6:58 PM
markj closed D33138: elf.5: Document compressed section headers.
Mon, Nov 29, 6:58 PM
markj closed D33133: dummynet: Fix socket option length validation for IP_DUMMYNET3.
Mon, Nov 29, 6:58 PM
markj closed D33132: dummynet: Avoid an out-of-bounds read in do_config().
Mon, Nov 29, 6:58 PM
markj added inline comments to D33161: bhyve: add ioctl to modify the second level address translation.
Mon, Nov 29, 5:48 PM · bhyve
markj added inline comments to D33129: bhyve: add ROM emulation.
Mon, Nov 29, 5:41 PM · bhyve
markj added a comment to D33011: bhyve: add ioctl to query infos about special memory regions.

I don't really like this interface. The name is too general and the interface is kind of simplistic, for example because it can only describe a single contiguous memory region and can't refer to different instances of a device. I'm not what a better interface would look like, I haven't yet read more of the patch series. But I think this deserves some more thought.

Mon, Nov 29, 5:28 PM · bhyve
markj accepted D33118: bhyve: add more slop to 64 bit BARs.
Mon, Nov 29, 5:04 PM · bhyve
markj added inline comments to D33165: swapoff(8): add -f.
Mon, Nov 29, 5:01 PM
markj accepted D32657: bhyve: Fix the wrong Number of Power States Support value also set a reasonable Max Power for NVMe controller.

Looks ok. I'd explicitly initialize MXPS and NOPS to zero as well.

Mon, Nov 29, 4:45 PM
markj accepted D31981: bhyve: allow reading of fwctl signature multiple times.
Mon, Nov 29, 4:36 PM · bhyve
markj added inline comments to D28278: bhyve: enumerate BARs by size.
Mon, Nov 29, 4:32 PM · bhyve
markj accepted D33147: Make swapoff more robust, when using swap into file.
Mon, Nov 29, 4:25 PM
markj added a comment to D15815: Set page permissions for kernel modules.

I believe this is superseded by commit 1d9eae9fb2e2253ca3d3764a5cc7f124b10e358b .

Mon, Nov 29, 4:07 PM
markj added a comment to D33154: e1000: fix interface capabilities management.

Could you please upload a diff with context?

Mon, Nov 29, 4:01 PM
markj accepted D33157: e1000: remove unused ifp backpointer.
Mon, Nov 29, 4:00 PM
markj added inline comments to D33147: Make swapoff more robust, when using swap into file.
Mon, Nov 29, 3:46 PM
markj accepted D33146: citrus_prop.c: do not ignore sign.
Mon, Nov 29, 3:37 PM
markj updated the diff for D33133: dummynet: Fix socket option length validation for IP_DUMMYNET3.

Make do_config() take a size_t instead.

Mon, Nov 29, 3:11 PM
markj added a comment to D33133: dummynet: Fix socket option length validation for IP_DUMMYNET3.
In D33133#748895, @kp wrote:

Wouldn't it make more sense to change l to be a size_t? sopt->sopt_valsize is a size_t, and the arguments to sootcopyin() are also size_t.
It's only do_config() which takes an int, and that may want to be changed to take a size_t as well.

Mon, Nov 29, 2:57 PM
markj updated the diff for D33139: libdwarf: Support consumption of compressed ELF sections.

size_t -> unsigned long

Mon, Nov 29, 2:44 PM
markj updated the diff for D33138: elf.5: Document compressed section headers.

a -> an
Use .Vt for referring to an Elf(32|64)_Chdr

Mon, Nov 29, 2:41 PM
markj committed R10:4022f3eec404: cddl: fix missing ZFS library dependencies (authored by greg_unrelenting.technology).
cddl: fix missing ZFS library dependencies
Mon, Nov 29, 2:29 PM
markj committed R10:d364adc09c05: Hoist cddl prebuild lib dependency definitions out of a MK_ZFS block (authored by markj).
Hoist cddl prebuild lib dependency definitions out of a MK_ZFS block
Mon, Nov 29, 2:28 PM
markj committed R10:f47775628333: Fix segment size in compressing core dumps (authored by jhibbits).
Fix segment size in compressing core dumps
Mon, Nov 29, 2:23 PM
markj committed R10:e0cee46f5800: amd64: Initialize kernel_pmap's active CPU set to all_cpus (authored by markj).
amd64: Initialize kernel_pmap's active CPU set to all_cpus
Mon, Nov 29, 2:22 PM
markj committed R10:1556ae13560c: vm_page: Remove vm_page_sbusy() and vm_page_xbusy() (authored by markj).
vm_page: Remove vm_page_sbusy() and vm_page_xbusy()
Mon, Nov 29, 2:22 PM
markj committed R10:cb081566cf86: vm_page: Consolidate page busy sleep mechanisms (authored by markj).
vm_page: Consolidate page busy sleep mechanisms
Mon, Nov 29, 2:22 PM
markj committed R10:294965542720: Fix segment size in compressing core dumps (authored by jhibbits).
Fix segment size in compressing core dumps
Mon, Nov 29, 2:22 PM
markj committed R10:fdd27db34802: vm: Add a mode to vm_object_page_remove() which skips invalid pages (authored by markj).
vm: Add a mode to vm_object_page_remove() which skips invalid pages
Mon, Nov 29, 2:22 PM
markj committed R10:141f0db23eff: pchtherm: Let the driver be compiled into the kernel (authored by N.J. Mann <njm@njm.me.uk>).
pchtherm: Let the driver be compiled into the kernel
Mon, Nov 29, 2:21 PM

Sat, Nov 27

markj added reviewers for D33139: libdwarf: Support consumption of compressed ELF sections: kaiw, jkoshy_users.sourceforge.net.
Sat, Nov 27, 1:30 AM
markj requested review of D33139: libdwarf: Support consumption of compressed ELF sections.
Sat, Nov 27, 1:30 AM
markj requested review of D33138: elf.5: Document compressed section headers.
Sat, Nov 27, 1:30 AM

Fri, Nov 26

markj requested review of D33133: dummynet: Fix socket option length validation for IP_DUMMYNET3.
Fri, Nov 26, 4:02 PM
markj requested review of D33132: dummynet: Avoid an out-of-bounds read in do_config().
Fri, Nov 26, 4:02 PM
markj committed R10:7b3642da21cb: Hoist cddl prebuild lib dependency definitions out of a MK_ZFS block (authored by markj).
Hoist cddl prebuild lib dependency definitions out of a MK_ZFS block
Fri, Nov 26, 2:40 PM

Thu, Nov 25

markj committed R10:b11e6fd75b1b: link_elf_obj: Process global ifunc relocs after other global relocs (authored by markj).
link_elf_obj: Process global ifunc relocs after other global relocs
Thu, Nov 25, 9:54 PM
markj closed D33120: link_elf_obj: Process global ifunc relocs after other global relocs.
Thu, Nov 25, 9:54 PM
markj added a comment to D33120: link_elf_obj: Process global ifunc relocs after other global relocs.
In D33120#748480, @kib wrote:

This is definitely an improvement because before, global ifunc symbols were not resolved at all.

But ultimately, it does not matter if the symbol is global or not, ifuncs generally should be processed after other types of relocations, global or local.

Thu, Nov 25, 8:23 PM
markj accepted D33119: swap on file: fix locking.
Thu, Nov 25, 7:11 PM
markj requested review of D33120: link_elf_obj: Process global ifunc relocs after other global relocs.
Thu, Nov 25, 6:53 PM
markj committed R10:cc7decfcf93b: make_dev.9: Fix a typo (authored by markj).
make_dev.9: Fix a typo
Thu, Nov 25, 4:38 PM
markj closed D33068: agp: Gracefully handle make_dev() errors if /dev/agpgart already exists.
Thu, Nov 25, 4:38 PM
markj committed R10:437ea82ce7fc: agp: Handle multiple devices more gracefully (authored by markj).
agp: Handle multiple devices more gracefully
Thu, Nov 25, 4:38 PM
markj added inline comments to D32498: Test various header files to ensure they can be included by themselves..
Thu, Nov 25, 2:22 PM
markj accepted D32498: Test various header files to ensure they can be included by themselves..

This looks good!

Thu, Nov 25, 2:21 PM

Wed, Nov 24

markj updated the diff for D33068: agp: Gracefully handle make_dev() errors if /dev/agpgart already exists.

Attempt to create the alias iff the device unit number is 0.

Wed, Nov 24, 7:09 PM
markj committed R10:d47d3a94bb2d: vm_fault: Factor out per-object operations into vm_fault_object() (authored by markj).
vm_fault: Factor out per-object operations into vm_fault_object()
Wed, Nov 24, 7:04 PM
markj committed R10:f1b642c255a2: vm_fault: Introduce a fault_status enum for internal return types (authored by markj).
vm_fault: Introduce a fault_status enum for internal return types
Wed, Nov 24, 7:04 PM
markj committed R10:45c09a74d681: vm_fault: Move nera into faultstate (authored by markj).
vm_fault: Move nera into faultstate
Wed, Nov 24, 7:03 PM
markj closed D33018: vm_fault: Factor out per-object operations into vm_fault_object().
Wed, Nov 24, 7:03 PM
markj closed D33017: vm_fault: Introduce a fault_status enum for internal return types.
Wed, Nov 24, 7:03 PM
markj closed D33016: vm_fault: Move nera into faultstate.
Wed, Nov 24, 7:03 PM
markj committed R10:517373f72302: natd: Remove uneeded in_cksum.h includes (authored by markj).
natd: Remove uneeded in_cksum.h includes
Wed, Nov 24, 6:40 PM
markj committed R10:60c95f316374: safexcel: Fix -Wunused-but-set-variable warnings (authored by markj).
safexcel: Fix -Wunused-but-set-variable warnings
Wed, Nov 24, 6:40 PM
markj committed R10:44775b163bfa: netinet: Remove unneeded mb_unmapped_to_ext() calls (authored by markj).
netinet: Remove unneeded mb_unmapped_to_ext() calls
Wed, Nov 24, 6:40 PM
markj committed R10:0d9c3423f59b: netinet: Implement in_cksum_skip() using m_apply() (authored by markj).
netinet: Implement in_cksum_skip() using m_apply()
Wed, Nov 24, 6:40 PM
markj committed R10:ecbbe831445a: netinet: Deduplicate most in_cksum() implementations (authored by markj).
netinet: Deduplicate most in_cksum() implementations
Wed, Nov 24, 6:40 PM
markj closed D33097: netinet: Remove unneeded mb_unmapped_to_ext() calls.
Wed, Nov 24, 6:40 PM
markj committed R10:5195bcc2126b: netinet: Remove in_cksum.c (authored by markj).
netinet: Remove in_cksum.c
Wed, Nov 24, 6:40 PM
markj closed D33096: netinet: Implement in_cksum_skip() using m_apply().
Wed, Nov 24, 6:40 PM
markj closed D33095: netinet: Deduplicate most in_cksum() implementations.
Wed, Nov 24, 6:40 PM
markj committed R10:09100f936be0: netinet: Remove in_cksum_update() (authored by markj).
netinet: Remove in_cksum_update()
Wed, Nov 24, 6:40 PM
markj closed D33094: netinet: Remove in_cksum.c.
Wed, Nov 24, 6:40 PM
markj closed D33093: netinet: Remove in_cksum_update().
Wed, Nov 24, 6:39 PM
markj added a comment to D33068: agp: Gracefully handle make_dev() errors if /dev/agpgart already exists.
In D33068#747710, @imp wrote:
In D33068#747080, @kib wrote:

IMO if we pretend that the driver is useful, it should create units as mentioned in summary. Or disable attach of any but first unit.

What is the canonical way to disable attach of all but the first unit? Is there some newbus mechanism for this?

I think 'if (device_get_unit(dev) != 0) return ENXIO;' in the attach routine. That produces an ugly error though,
so often times I've seen in private code 'return 0' and have no resources created for the device, perhaps apart
from activating the resources automatically assigned to the device.

Wed, Nov 24, 5:14 PM
markj updated the diff for D33068: agp: Gracefully handle make_dev() errors if /dev/agpgart already exists.

Go a bit further and create /dev/agpgart%d for all devices, and create the alias
for the device with unit number 0.

Wed, Nov 24, 5:12 PM
markj committed R10:f69572733023: Export symbols from opensolaris.ko and dtrace.ko (authored by markj).
Export symbols from opensolaris.ko and dtrace.ko
Wed, Nov 24, 3:56 PM

Tue, Nov 23

markj updated the diff for D33097: netinet: Remove unneeded mb_unmapped_to_ext() calls.
  • Fix ip_output()
  • Move the mb_unmapped_to_ext() call out of ip6_output_delayed_csum()
Tue, Nov 23, 6:46 PM
markj added inline comments to D33097: netinet: Remove unneeded mb_unmapped_to_ext() calls.
Tue, Nov 23, 6:45 PM
markj added a comment to D33095: netinet: Deduplicate most in_cksum() implementations.

P.S., I spent some time trying to replace this implementation with an updated one from NetBSD. Theirs is easier to read and is licensed under BSD 2-clause. It got somewhat messy, though, and wasn't important to my immediate goal, so I dropped it. I think there's also some room for micro-optimizations here, for instance in_pseudo() and in_addword() should probably be inlines. in_pseudo() in particular is dead simple, it's just a ones complement sum:

Tue, Nov 23, 4:51 PM
markj added inline comments to D33097: netinet: Remove unneeded mb_unmapped_to_ext() calls.
Tue, Nov 23, 4:24 PM
markj added a comment to D33068: agp: Gracefully handle make_dev() errors if /dev/agpgart already exists.
In D33068#747080, @kib wrote:

IMO if we pretend that the driver is useful, it should create units as mentioned in summary. Or disable attach of any but first unit.

Tue, Nov 23, 2:56 PM
markj committed R10:83d0a7763a92: aesni: Avoid a potential out-of-bounds load in AES_GCM_encrypt() (authored by markj).
aesni: Avoid a potential out-of-bounds load in AES_GCM_encrypt()
Tue, Nov 23, 2:35 PM
markj committed R10:35dfdb88eafd: unix: Remove a write-only local variable (authored by markj).
unix: Remove a write-only local variable
Tue, Nov 23, 2:34 PM
markj committed R10:8c29b0eeb0ce: aesni: Avoid a potential out-of-bounds load in AES_GCM_encrypt() (authored by markj).
aesni: Avoid a potential out-of-bounds load in AES_GCM_encrypt()
Tue, Nov 23, 2:33 PM
markj added reviewers for D33097: netinet: Remove unneeded mb_unmapped_to_ext() calls: jhb, gallatin, glebius.
Tue, Nov 23, 2:31 PM
markj added reviewers for D33096: netinet: Implement in_cksum_skip() using m_apply(): jhb, gallatin, glebius.
Tue, Nov 23, 2:31 PM
markj requested review of D33097: netinet: Remove unneeded mb_unmapped_to_ext() calls.
Tue, Nov 23, 2:30 PM
markj requested review of D33096: netinet: Implement in_cksum_skip() using m_apply().
Tue, Nov 23, 2:30 PM
markj requested review of D33095: netinet: Deduplicate most in_cksum() implementations.
Tue, Nov 23, 2:30 PM
markj requested review of D33094: netinet: Remove in_cksum.c.
Tue, Nov 23, 2:29 PM
markj requested review of D33093: netinet: Remove in_cksum_update().
Tue, Nov 23, 2:29 PM

Mon, Nov 22

markj committed R10:a11983366ea7: pchtherm: Let the driver be compiled into the kernel (authored by N.J. Mann <njm@njm.me.uk>).
pchtherm: Let the driver be compiled into the kernel
Mon, Nov 22, 5:13 PM
markj committed R10:22082f15f9bb: When copying types from one CTF container to another, ensure that we (authored by jtl).
When copying types from one CTF container to another, ensure that we
Mon, Nov 22, 4:35 PM
markj committed R10:68396709e73a: libctf: Improve check for duplicate SOU definitions in ctf_add_type() (authored by markj).
libctf: Improve check for duplicate SOU definitions in ctf_add_type()
Mon, Nov 22, 4:35 PM
markj committed R10:4faff19d6305: When copying types from one CTF container to another, ensure that we (authored by jtl).
When copying types from one CTF container to another, ensure that we
Mon, Nov 22, 4:03 PM
markj added a comment to D31981: bhyve: allow reading of fwctl signature multiple times.

Is it sufficient to instead set be_state = IDENT_WAIT in fwctl_inl(), or handle be_state == REQ the same as be_state == IDENT_WAIT in fwctl_outw()?

Mon, Nov 22, 2:33 PM · bhyve
markj accepted D31051: bhyve: use a fixed 32 bit BAR base address.
Mon, Nov 22, 2:16 PM · bhyve
markj accepted D28278: bhyve: enumerate BARs by size.

I think D27220 was intended to solve the same problem? I'm not sure if a bhyve patch to use libuvmem is available.

This seems like a reasonable interim solution.

I'm unfamiliar with libuvmem. I'm unsure whether libuvmem is able to solve the problem too or not.

Mon, Nov 22, 2:07 PM · bhyve