Page MenuHomeFreeBSD

mhorne (Mitchell Horne)
User

Projects

User Details

User Since
Mar 22 2019, 4:46 AM (95 w, 2 d)

Recent Activity

Fri, Jan 15

mhorne added a comment to D28180: bsdinstall: riscv platform tweaks.

Note that partedit_riscv.c is practically identical to partedit_arm64.c. We could merge the two, I'm just not sure what it ought to be called.

Fri, Jan 15, 5:53 PM
mhorne requested review of D28180: bsdinstall: riscv platform tweaks.
Fri, Jan 15, 5:50 PM
mhorne requested review of D28176: bsdinstall: create /EFI/BOOT directory in ESP.
Fri, Jan 15, 4:19 PM
mhorne committed R10:0b92d1dd18c1: riscv: fix kernel build (authored by mhorne).
riscv: fix kernel build
Fri, Jan 15, 4:12 PM

Thu, Jan 14

mhorne added a comment to D28151: risc-v : fix dtrace-check for fp == oldfp.

Hi, there is an open review in D28054, ......

Hi, ah, o.k., didn`t see that. I didn`t make further dtrace-tests because still struggling with
solving the bootup-panic of the HiFiveUnleashed.

Thu, Jan 14, 3:34 PM
mhorne abandoned D28104: allwinner: simplify axp81x read/write.
In D28104#629208, @manu wrote:

Yeah I remember now.
RSB isn't really i2c and we can't use those (I was sure that I tested at some point).
using this code will boot but will fail as soon as we try to change the cpu freq (when we try to talk to the PMIC)
root@pine64-lts:~ # sysctl dev.cpu.0.freq=1152
dev.cpu.0.freq: iichb0: RTA not known for address 0x23
axp8xx_pmu0: Setting vdd-cpux (dcdc2) to 1300000<->1300000
<hang here>
Since we cannot communicate with the PMIC we will hang as soon as the CPU reaches the desired freq and not have the correct needed voltage.

Thu, Jan 14, 2:45 PM
mhorne added a comment to D28151: risc-v : fix dtrace-check for fp == oldfp.

Hi, there is an open review in D28054, discussing the issues with this function (and included build fix).

Thu, Jan 14, 2:42 PM
mhorne added a comment to D28054: dtrace: Fix RISC-V user stack unwinder.

Looking at other implementations, I believe oldfp should be updated on each loop iteration.

Thu, Jan 14, 2:40 PM

Wed, Jan 13

mhorne committed R10:818390ce0ca5: arm64: fix early devmap assertion (authored by mhorne).
arm64: fix early devmap assertion
Wed, Jan 13, 9:28 PM
mhorne accepted D27844: make maximum interrupt number tunable on ARM, ARM64, MIPS, and RISC-V.

A couple of small nitpicks, otherwise LGTM.

Wed, Jan 13, 8:04 PM
mhorne added a comment to D28126: arm64: adjust bswap inline assembly constraints.
In D28126#628610, @mjg wrote:

This (like everything else) should use compiler instrinsics by default, see __builtin_bswap32 for example.

Wed, Jan 13, 6:24 PM
mhorne added a comment to D28104: allwinner: simplify axp81x read/write.
In D28104#628657, @manu wrote:

Mhm, I don't remember if there was a reason I didn't do that or if iicdev_read/write didn't exists at the time.
Looks ok but give me a few days to test.

Wed, Jan 13, 2:53 PM

Tue, Jan 12

mhorne requested review of D28127: ddb: fix show devmap output on 32-bit arm.
Tue, Jan 12, 10:38 PM
mhorne requested review of D28126: arm64: adjust bswap inline assembly constraints.
Tue, Jan 12, 10:34 PM
mhorne committed R10:d89e1db5a331: if_wg: fix modules load on !x86 (authored by mhorne).
if_wg: fix modules load on !x86
Tue, Jan 12, 10:08 PM
mhorne committed R10:0628f6835702: riscv pmap: add some pv list assertions (authored by mhorne).
riscv pmap: add some pv list assertions
Tue, Jan 12, 3:16 PM
mhorne closed D28109: riscv pmap: add a demotion assertion.
Tue, Jan 12, 3:15 PM
mhorne updated the diff for D28109: riscv pmap: add a demotion assertion.

Add asserts to pmap_page_test_mappings() and pmap_page_wired_mappings().

Tue, Jan 12, 2:07 PM
mhorne added a comment to D28109: riscv pmap: add a demotion assertion.

I think pmap_page_test_mappings() and pmap_page_wired_mappings() could use a similar treatment.

Tue, Jan 12, 1:57 AM
mhorne added inline comments to D28109: riscv pmap: add a demotion assertion.
Tue, Jan 12, 1:50 AM
mhorne requested review of D28109: riscv pmap: add a demotion assertion.
Tue, Jan 12, 1:43 AM

Mon, Jan 11

mhorne accepted D28098: arm64: Don't disable options GDB in LINT..

Thanks for catching this!

Mon, Jan 11, 10:12 PM
mhorne requested review of D28104: allwinner: simplify axp81x read/write.
Mon, Jan 11, 10:08 PM
mhorne accepted D28100: Enable accelerated AES-XTS software crypto in GENERIC..

Would there be any benefit (or risk) to enabling ossl(4) by default as well?

Mon, Jan 11, 8:11 PM
mhorne accepted D28099: arm64: Add armv8crpyto and ossl to NOTES..
Mon, Jan 11, 8:06 PM
mhorne requested changes to D28098: arm64: Don't disable options GDB in LINT..

I see options GDB in the common NOTES already.

Mon, Jan 11, 8:01 PM
mhorne added a comment to D28095: gitignore: expand list of ignored files.

Note that I have D28047 uncommitted for cscope files, but am happy to abandon that in favor of this. I also don't see a need for ?cscope.out.

Mon, Jan 11, 6:32 PM
mhorne committed R10:7dcddacafaf0: cgem: update 64-bit check (authored by mhorne).
cgem: update 64-bit check
Mon, Jan 11, 4:24 PM

Sun, Jan 10

mhorne closed D28048: rc.conf(5): describe devmatch rc variables.
Sun, Jan 10, 8:58 PM
mhorne committed R10:ef757da441b1: rc.conf(5): describe devmatch rc variables (authored by mhorne).
rc.conf(5): describe devmatch rc variables
Sun, Jan 10, 8:58 PM
mhorne closed D24304: Add 64-bit support to Cadence CGEM Ethernet driver..
Sun, Jan 10, 8:54 PM
mhorne committed R10:facdd1cd2045: cgem: add 64-bit support (authored by skibo).
cgem: add 64-bit support
Sun, Jan 10, 8:54 PM

Sat, Jan 9

mhorne added a comment to D24304: Add 64-bit support to Cadence CGEM Ethernet driver..

Makefile update looks good to me! You might want to bump .Dd when you commit.

I am not a committer but I bumped the date. Earlier today, I sanity checked the driver on both Zynq board and a Zynq UltraScale so I think it's ready.

Sat, Jan 9, 12:21 AM

Fri, Jan 8

mhorne added inline comments to D28048: rc.conf(5): describe devmatch rc variables.
Fri, Jan 8, 11:20 PM
mhorne updated the diff for D28048: rc.conf(5): describe devmatch rc variables.

Use the full month.

Fri, Jan 8, 11:19 PM
mhorne committed R10:cbc9be948a3b: sifive_uart: quiet GCC -Werror=parentheses (authored by mhorne).
sifive_uart: quiet GCC -Werror=parentheses
Fri, Jan 8, 9:37 PM
mhorne requested review of D28048: rc.conf(5): describe devmatch rc variables.
Fri, Jan 8, 8:58 PM
mhorne added a comment to D28047: gitignore: update cscope rules.

@khng300_gmail.com am I right in thinking you have one or two outstanding .gitignore reviews?

Fri, Jan 8, 8:49 PM
mhorne requested review of D28047: gitignore: update cscope rules.
Fri, Jan 8, 8:48 PM
mhorne requested review of D28046: gdb: only return signal values for powerpc's gdb_cpu_signal().
Fri, Jan 8, 8:31 PM
mhorne added inline comments to D27719: arm64: don't pass user trapframe to kdb_trap().
Fri, Jan 8, 7:16 PM
mhorne committed R10:bbfa199cbc16: arm64: gdb(4) machine-dependent bits (authored by mhorne).
arm64: gdb(4) machine-dependent bits
Fri, Jan 8, 6:54 PM
mhorne committed R10:e9bb4ce3d0e7: arm64: don't pass user trapframe to kdb_trap() (authored by mhorne).
arm64: don't pass user trapframe to kdb_trap()
Fri, Jan 8, 6:54 PM
mhorne committed R10:5f66d5a313bf: arm64: remove pcb_pc (authored by mhorne).
arm64: remove pcb_pc
Fri, Jan 8, 6:54 PM
mhorne committed R10:088a7eef95b1: ddb: add ability to print user registers (authored by mhorne).
ddb: add ability to print user registers
Fri, Jan 8, 6:54 PM
mhorne closed D27727: arm64: gdb(4) stub.
Fri, Jan 8, 6:54 PM
mhorne closed D27720: arm64: remove pcb_pc.
Fri, Jan 8, 6:54 PM
mhorne closed D27705: ddb: add ability to print user registers.
Fri, Jan 8, 6:54 PM
mhorne added a comment to D24304: Add 64-bit support to Cadence CGEM Ethernet driver..

Makefile update looks good to me! You might want to bump .Dd when you commit.

Fri, Jan 8, 4:14 PM
mhorne accepted D26991: Optimize set_syscall_retval for arm64.

See also, the analogous commit for riscv: rS367138.

Fri, Jan 8, 3:59 PM

Thu, Jan 7

mhorne added a comment to D24304: Add 64-bit support to Cadence CGEM Ethernet driver..

Hi @skibo , is there any remaining work to be done on this patch? It appears to be ready to go, other than one small nitpick. It would be nice to see it committed before stable/13 branches later this month.

Thu, Jan 7, 8:21 PM
mhorne committed R10:40d0971bbe5d: arm64: enable build of the ipmi module (authored by mhorne).
arm64: enable build of the ipmi module
Thu, Jan 7, 7:48 PM
mhorne closed D28009: WIP: arm64: enable the ipmi module.
Thu, Jan 7, 7:48 PM
mhorne closed D21017: armv8crypto: add AES-XTS support.
Thu, Jan 7, 7:37 PM · arm64
mhorne committed R10:4979620ece98: armv8crypto: add AES-XTS support (authored by mhorne).
armv8crypto: add AES-XTS support
Thu, Jan 7, 7:37 PM
mhorne updated the diff for D27727: arm64: gdb(4) stub.

Remove duplicate function prototypes. Small changes to formatting and comments.

Thu, Jan 7, 6:48 PM
mhorne added a comment to D28009: WIP: arm64: enable the ipmi module.

I wonder if there are machines that support any of the existing communication methods. eMAG would need acpi-i2c support…

Thu, Jan 7, 5:14 PM
mhorne updated the diff for D28009: WIP: arm64: enable the ipmi module.

Reorder SRCS blocks in ipmi/Makefile. Set _ipmi in arm64/amd64/i386 block (powerpc64 block untouched).

Thu, Jan 7, 5:07 PM
mhorne added inline comments to D27727: arm64: gdb(4) stub.
Thu, Jan 7, 3:59 PM

Wed, Jan 6

mhorne retitled D28009: WIP: arm64: enable the ipmi module from arm64: enable the ipmi module to WIP: arm64: enable the ipmi module.
Wed, Jan 6, 9:05 PM
mhorne requested review of D28009: WIP: arm64: enable the ipmi module.
Wed, Jan 6, 8:31 PM
mhorne added inline comments to D28001: arm64 GENERIC: sort to match amd64, to ease comparison.
Wed, Jan 6, 6:06 PM
mhorne added inline comments to D27727: arm64: gdb(4) stub.
Wed, Jan 6, 3:18 PM

Tue, Jan 5

mhorne updated the diff for D27719: arm64: don't pass user trapframe to kdb_trap().

Remove the tf_elr manipulation.

Tue, Jan 5, 9:39 PM

Mon, Jan 4

mhorne committed R10:6a45b9926f2e: gdb(4) fix x86 signal reporting (authored by mhorne).
gdb(4) fix x86 signal reporting
Mon, Jan 4, 9:12 PM
mhorne committed R10:1c002413c3d5: gdb(4): allow bulk write of registers (authored by mhorne).
gdb(4): allow bulk write of registers
Mon, Jan 4, 9:12 PM
mhorne committed R10:7b7138454fd3: gdb(4): handle single register read packets (authored by mhorne).
gdb(4): handle single register read packets
Mon, Jan 4, 9:12 PM
mhorne committed R10:aba250f6264f: amd64: use register macros for gdb_cpu_getreg() (authored by mhorne).
amd64: use register macros for gdb_cpu_getreg()
Mon, Jan 4, 9:12 PM
mhorne committed R10:cfe79bddf4b0: amd64: allow gdb(4) to write to most registers (authored by mhorne).
amd64: allow gdb(4) to write to most registers
Mon, Jan 4, 9:12 PM
mhorne added a comment to D27454: Add AES-GCM H/W acceleration for kTLS on ARMv8 architecture.

I'm looking at D21017, which adds AES-XTS support to armv8_crypto, and has been sitting in review for some time.

Mon, Jan 4, 7:31 PM
mhorne added inline comments to D27844: make maximum interrupt number tunable on ARM, ARM64, MIPS, and RISC-V.
Mon, Jan 4, 7:03 PM
mhorne added a comment to D27720: arm64: remove pcb_pc.
In D27720#622405, @jhb wrote:

For KGDB I explicitly ignore pcb_pc, but I do assume the layout for getting to the pc_sp below using the following structure:

static const struct regcache_map_entry aarch64_fbsd_pcbmap[] =
  {
    { 30, AARCH64_X0_REGNUM, 8 }, /* x0 ... x29 */
    { 1, AARCH64_PC_REGNUM, 8 },
    { 1, REGCACHE_MAP_SKIP, 8 },
    { 1, AARCH64_SP_REGNUM, 8 },
    { 0 }
  };

I could perhaps depend on osreldate which holds __FreeBSD_version to handle the differing layouts by having two versions of the structure.

Mon, Jan 4, 3:47 PM
mhorne updated the diff for D27705: ddb: add ability to print user registers.

Fix typo, kdb_frame -> kdb_thread. Make it a little clearer that this is the
previous trapframe and not necissarily the user trapframe.

Mon, Jan 4, 2:49 PM

Sat, Jan 2

mhorne closed D27644: gdb(4): allow bulk write and single read of registers.
Sat, Jan 2, 11:09 PM
mhorne closed D27645: gdb(4): fix x86 signal reporting.
Sat, Jan 2, 11:08 PM
mhorne closed D27535: Handle return from KDB_WHY_TRAP on arm and mips.
Sat, Jan 2, 11:07 PM
mhorne committed R10:65d1a4cfda73: Handle KDB_WHY_TRAP return on arm and mips (authored by mhorne).
Handle KDB_WHY_TRAP return on arm and mips
Sat, Jan 2, 11:07 PM

Wed, Dec 23

mhorne added inline comments to D27720: arm64: remove pcb_pc.
Wed, Dec 23, 8:36 PM

Tue, Dec 22

mhorne added a comment to D27645: gdb(4): fix x86 signal reporting.

On a second look, I found that gdb_cpu_signal() only returns standard POSIX signals, meaning that at present no mapping is required between FreeBSD and GDB signals. PowerPC is the slight exception to this, which might return a trap vector instead of a signal, and will be addressed.

Tue, Dec 22, 7:33 PM
mhorne updated the diff for D27644: gdb(4): allow bulk write and single read of registers.

Don't transmit "ok" for the error case.

Tue, Dec 22, 7:17 PM
mhorne added a comment to D21017: armv8crypto: add AES-XTS support.

I was able to rebase this patch with some very minor tweaks, and verify the results using both cryptocheck and cryptotest.py.

Tue, Dec 22, 6:27 PM · arm64
mhorne updated the summary of D27727: arm64: gdb(4) stub.
Tue, Dec 22, 6:09 PM
mhorne added a comment to D27727: arm64: gdb(4) stub.

I will be on holiday for the next couple weeks, but posting this now in case anyone wants to look it over during that time. No rush to review.

Tue, Dec 22, 6:08 PM
mhorne retitled D27727: arm64: gdb(4) stub from Summary: to arm64: gdb(4) stub.
Tue, Dec 22, 6:06 PM
mhorne requested review of D27727: arm64: gdb(4) stub.
Tue, Dec 22, 6:01 PM
mhorne requested review of D27720: arm64: remove pcb_pc.
Tue, Dec 22, 2:58 PM
mhorne requested review of D27719: arm64: don't pass user trapframe to kdb_trap().
Tue, Dec 22, 2:30 PM

Mon, Dec 21

mhorne requested review of D27705: ddb: add ability to print user registers.
Mon, Dec 21, 8:52 PM

Fri, Dec 18

mhorne closed D27660: riscv: report additional known SBI implementations.
Fri, Dec 18, 8:10 PM
mhorne committed rS368767: riscv: report additional known SBI implementations.
riscv: report additional known SBI implementations
Fri, Dec 18, 8:10 PM

Dec 18 2020

mhorne committed rS368765: amd64: use register macros for gdb_cpu_getreg().
amd64: use register macros for gdb_cpu_getreg()
Dec 18 2020, 4:16 PM
mhorne closed D27642: amd64: allow gdb(4) to write to most registers.
Dec 18 2020, 4:10 PM
mhorne committed rS368764: amd64: allow gdb(4) to write to most registers.
amd64: allow gdb(4) to write to most registers
Dec 18 2020, 4:10 PM

Dec 17 2020

mhorne added inline comments to D27642: amd64: allow gdb(4) to write to most registers.
Dec 17 2020, 9:43 PM
mhorne requested review of D27660: riscv: report additional known SBI implementations.
Dec 17 2020, 8:49 PM
mhorne updated the diff for D27642: amd64: allow gdb(4) to write to most registers.

Fix typo, GDB_REG_RIP -> GDB_REG_PC.

Dec 17 2020, 8:14 PM
mhorne updated the diff for D27642: amd64: allow gdb(4) to write to most registers.

Update to use two switch statements. This version fits vertically on a single screen, and more closely matches the formatting of gdb_cpu_setreg().

Dec 17 2020, 8:10 PM
mhorne added a comment to D27645: gdb(4): fix x86 signal reporting.
In D27645#618106, @jhb wrote:

FWIW, in the bhyve stub, I just always return '5' (SIGTRAP). Note that the signal numbers returned are defined in GDB's source and aren't necessarily FreeBSD signal number values. See the last sentence in the first paragraph here:

https://sourceware.org/gdb/current/onlinedocs/gdb/Stop-Reply-Packets.html#Stop-Reply-Packets

The actual values are in signals.def:

https://github.com/bsdjhb/gdb/blob/master/include/gdb/signals.def.

Not sure if in gdb(4) we are using some kind of translation table around gdb_cpu_signal to map FreeBSD signal numbers to GDB signal numbers, but that is probably the best way to fix that if needed. For userland we do this here:

https://github.com/bsdjhb/gdb/blob/master/gdb/fbsd-tdep.c#L1727

Dec 17 2020, 7:49 PM
mhorne added a comment to D27644: gdb(4): allow bulk write and single read of registers.
In D27644#618098, @jhb wrote:

I might split these into two. I suspect only 'p' is actually needed for call, but it doesn't hurt to have G I guess?

Dec 17 2020, 7:21 PM
mhorne closed D25737: RISC-V u-boot ports.
Dec 17 2020, 6:12 PM