wma (Wojciech Macek)
User

Projects

User Details

User Since
Jan 20 2016, 6:04 AM (104 w, 1 d)

Recent Activity

Today

wma committed rS328113: Call platform_smp_ap_init before decr_ap_init.
Call platform_smp_ap_init before decr_ap_init
Thu, Jan 18, 8:34 AM
wma closed D13947: PPC64: fix TOC behavior on process initialization.
Thu, Jan 18, 7:43 AM
wma committed rS328112: PPC64: fix TOC behavior on process initialization.
PPC64: fix TOC behavior on process initialization
Thu, Jan 18, 7:43 AM
wma committed rS328111: PPC: machdep, zero BSS always but BookE.
PPC: machdep, zero BSS always but BookE
Thu, Jan 18, 7:41 AM
wma closed D13948: PPC: machdep, zero BSS always but BookE.
Thu, Jan 18, 7:41 AM
wma closed D13949: KDB: restart only CPUs stopped by KDB.
Thu, Jan 18, 7:39 AM
wma committed rS328110: KDB: restart only CPUs stopped by KDB.
KDB: restart only CPUs stopped by KDB
Thu, Jan 18, 7:39 AM
wma committed rS328109: PPC64: add AHCI back to GENERIC64.
PPC64: add AHCI back to GENERIC64
Thu, Jan 18, 6:28 AM

Yesterday

wma created D13950: IFLIB: do not remove dmamap on buffer unload.
Wed, Jan 17, 11:42 AM
wma created D13949: KDB: restart only CPUs stopped by KDB.
Wed, Jan 17, 10:22 AM
wma created D13948: PPC: machdep, zero BSS always but BookE.
Wed, Jan 17, 10:20 AM
wma created D13947: PPC64: fix TOC behavior on process initialization.
Wed, Jan 17, 10:09 AM
wma committed rS328080: PPC64: implement missing busdma ops.
PPC64: implement missing busdma ops
Wed, Jan 17, 9:45 AM
wma committed rS328079: PPC64: fix copyinout ranges.
PPC64: fix copyinout ranges
Wed, Jan 17, 9:36 AM
wma committed rS328078: PPC64: add CXGBE and remove AHCI from GENERIC64.
PPC64: add CXGBE and remove AHCI from GENERIC64
Wed, Jan 17, 9:33 AM
wma committed rS328077: PowerNV: workaround console on OPAL 5.4.
PowerNV: workaround console on OPAL 5.4
Wed, Jan 17, 8:02 AM
wma committed rS328076: PowerNV: make PowerNV PCIe working on a real hardware.
PowerNV: make PowerNV PCIe working on a real hardware
Wed, Jan 17, 7:39 AM
wma abandoned D13883: Use appropriate alternative to -msoft-float for clang-built PowerPC kernels..

Yes, the recent toolchain works well.

Wed, Jan 17, 6:46 AM

Tue, Jan 16

wma abandoned D6336: Disable attaching of generic PCI drv on FDT nodes.

It's already outdated

Tue, Jan 16, 5:57 PM
wma abandoned D6986: ARM, ARM64: Workaround for buf_ring reordering.

Merged as r302292

Tue, Jan 16, 5:56 PM
wma created D13932: ULE: provide defaults to ts_cpu.
Tue, Jan 16, 7:53 AM
wma closed D13857: PowerNV: XICS support for PowerNV/OPAL.

Closed by
https://svnweb.freebsd.org/changeset/base/328042

Tue, Jan 16, 6:25 AM
wma committed rS328042: PowerNV: XICS support for PowerNV/OPAL.
PowerNV: XICS support for PowerNV/OPAL
Tue, Jan 16, 6:24 AM

Mon, Jan 15

wma updated the diff for D13857: PowerNV: XICS support for PowerNV/OPAL.
Mon, Jan 15, 11:02 AM

Fri, Jan 12

wma committed rS327873: PowerNV: update OPAL driver.
PowerNV: update OPAL driver
Fri, Jan 12, 12:15 PM
wma created D13883: Use appropriate alternative to -msoft-float for clang-built PowerPC kernels..
Fri, Jan 12, 11:48 AM
wma added a reviewer for D13857: PowerNV: XICS support for PowerNV/OPAL: mw.
Fri, Jan 12, 10:17 AM
wma added a comment to D13857: PowerNV: XICS support for PowerNV/OPAL.

Please let me know if there are any objections to this patch. I'd like to have it commited on Monday.

Fri, Jan 12, 10:16 AM
wma committed rS327867: PowerNV: initial support for PCIe host controller.
PowerNV: initial support for PCIe host controller
Fri, Jan 12, 7:56 AM

Thu, Jan 11

wma committed rS327815: PowerNV: add buffer for OPAL console.
PowerNV: add buffer for OPAL console
Thu, Jan 11, 9:42 AM
wma committed rS327814: PowerNV: set LPCR[LPES] correctly.
PowerNV: set LPCR[LPES] correctly
Thu, Jan 11, 9:39 AM
wma committed rS327813: PowerNV: correctly start secondary CPUs.
PowerNV: correctly start secondary CPUs
Thu, Jan 11, 9:34 AM
wma committed rS327812: PowerNV: add reset, poweroff, OPAL console.
PowerNV: add reset, poweroff, OPAL console
Thu, Jan 11, 9:26 AM
wma committed rS327811: CXGBE: fix get_filt to be endianness-aware.
CXGBE: fix get_filt to be endianness-aware
Thu, Jan 11, 9:17 AM
wma closed D13102: CXGBE: fix get_filt to be endianness-aware.
Thu, Jan 11, 9:17 AM
wma created D13857: PowerNV: XICS support for PowerNV/OPAL.
Thu, Jan 11, 8:22 AM
wma committed rS327810: PowerNV: initial support for OPAL.
PowerNV: initial support for OPAL
Thu, Jan 11, 7:40 AM

Tue, Jan 9

wma added a comment to D13102: CXGBE: fix get_filt to be endianness-aware.

Just tested on arm64 (LE), works fine.

Tue, Jan 9, 3:19 PM

Nov 16 2017

wma abandoned D12600: POWERPC: don't wait in loop if AP was not started.

Closing this one as I'm going to change this function much more.

Nov 16 2017, 10:49 AM

Nov 15 2017

wma created D13102: CXGBE: fix get_filt to be endianness-aware.
Nov 15 2017, 12:55 PM
wma closed D13084: CXGBE: fix big-endian behaviour.
Nov 15 2017, 6:46 AM
wma committed rS325840: CXGBE: fix big-endian behaviour.
CXGBE: fix big-endian behaviour
Nov 15 2017, 6:46 AM
wma added a comment to D13084: CXGBE: fix big-endian behaviour.

Yes, sysctl shows all doorbells just fine

Nov 15 2017, 6:45 AM

Nov 14 2017

wma created D13084: CXGBE: fix big-endian behaviour.
Nov 14 2017, 1:18 PM

Nov 2 2017

wma added a comment to D12600: POWERPC: don't wait in loop if AP was not started.

After some testing of all this CPU startup thing, I think it would be best to panic if any of AP fails to start.
The rest of the system is not prepared to handle the scenario when the cpumask has "holes" inside. I then can see hangs in various places: for example the smp_rendezvouz, returning from kdb_enter, but there still might be more such things which I couldn't find.

Nov 2 2017, 5:46 AM

Oct 6 2017

wma created D12600: POWERPC: don't wait in loop if AP was not started.
Oct 6 2017, 7:40 AM

Oct 2 2017

wma committed rS324187: PPC: increase MAX_PICS to 32.
PPC: increase MAX_PICS to 32
Oct 2 2017, 6:05 AM
wma closed D12540: PPC: increase MAX_PICS to 32.
Oct 2 2017, 6:05 AM

Sep 29 2017

wma updated the diff for D12422: loader: support for mixed-endianness ELF/loader and POWER8.
Sep 29 2017, 7:42 AM
wma abandoned D5991: Improve stack guard page handling on ARM.

Obsoleted

Sep 29 2017, 7:41 AM
wma added a reviewer for D12540: PPC: increase MAX_PICS to 32: mw.
Sep 29 2017, 7:31 AM
wma created D12540: PPC: increase MAX_PICS to 32.
Sep 29 2017, 6:41 AM
wma committed rS324099: Compile loader as Little-Endian on PPC64/POWER8.
Compile loader as Little-Endian on PPC64/POWER8
Sep 29 2017, 6:36 AM
wma closed D12421: Compile loader as Little-Endian on PPC64/POWER8.
Sep 29 2017, 6:36 AM

Sep 22 2017

wma added inline comments to D12421: Compile loader as Little-Endian on PPC64/POWER8.
Sep 22 2017, 11:33 AM
wma added a comment to D12421: Compile loader as Little-Endian on PPC64/POWER8.

Oh, I indeed forgot to add -U9999 to the format-patch, sorry.

Sep 22 2017, 11:30 AM
wma updated the diff for D12421: Compile loader as Little-Endian on PPC64/POWER8.
Sep 22 2017, 11:13 AM
wma added a comment to D12422: loader: support for mixed-endianness ELF/loader and POWER8.

If I understood you correctly, you're referring to the part such as load_elf, lines 131-177.
The purpose of this code is to read the ELF header and store it as ehdr structure, which can be accessed by CPU. To do that, following algorithm is used:

  1. Read the ELF header structure and store it in the memory. Loader operates on blocks of data, so we're loading the whole chunk of data at once using "block read" operation.
  2. We might fall into sutuation, that some ELF header contents are BE-encoded while the loader runs as LE.
  3. An exact ehdr structure shape denepends on the ELF arch/bits/etc. and is generated during compilation.
Sep 22 2017, 11:04 AM

Sep 21 2017

wma added a comment to D12422: loader: support for mixed-endianness ELF/loader and POWER8.

Kevin, is there any explanation for this change?
All ELF parsing code is machine-independent. Some architectures don't have instructions allowing choose of data endianness, so I don't see a way how could we avoid using generic byte-swap macros.
I've used instructions you mentioned in the machine-specific trampoline code, which is a part of ppc64-kernel bootstrap and where we no longer care about code interoperatibility.

Sep 21 2017, 8:10 AM

Sep 20 2017

wma added inline comments to D12422: loader: support for mixed-endianness ELF/loader and POWER8.
Sep 20 2017, 2:59 PM
wma added inline comments to D12422: loader: support for mixed-endianness ELF/loader and POWER8.
Sep 20 2017, 2:53 PM
wma added a comment to D12421: Compile loader as Little-Endian on PPC64/POWER8.

The Makefile.inc adds the -mlittle_endian flag to all dependencies automatically, including libstand32.

Sep 20 2017, 2:48 PM
wma created D12422: loader: support for mixed-endianness ELF/loader and POWER8.
Sep 20 2017, 11:09 AM
wma created D12421: Compile loader as Little-Endian on PPC64/POWER8.
Sep 20 2017, 11:05 AM

May 29 2017

wma committed rS319117: Introduce Genesys GL3224 quirks.
Introduce Genesys GL3224 quirks
May 29 2017, 9:23 AM
wma closed D10902: Introduce Genesys GL3224 quirks by committing rS319117: Introduce Genesys GL3224 quirks.
May 29 2017, 9:23 AM · ARM
wma closed D10903: Increase timeout in Atheros HAL by committing rS319116: Increase timeout in Atheros HAL.
May 29 2017, 9:22 AM · network
wma committed rS319116: Increase timeout in Atheros HAL.
Increase timeout in Atheros HAL
May 29 2017, 9:22 AM
wma committed rS319115: Enable wireless Atheros cards in ARMADA38X.
Enable wireless Atheros cards in ARMADA38X
May 29 2017, 9:20 AM
wma closed D10904: Enable wireless Atheros cards in ARMADA38X by committing rS319115: Enable wireless Atheros cards in ARMADA38X.
May 29 2017, 9:20 AM · ARM

May 19 2017

wma committed rS318526: Fix boot up on ARMADA38X uniprocessor variant.
Fix boot up on ARMADA38X uniprocessor variant
May 19 2017, 8:27 AM
wma closed D10783: Fix boot up on ARMADA38X uniprocessor variant by committing rS318526: Fix boot up on ARMADA38X uniprocessor variant.
May 19 2017, 8:27 AM · ARM
wma committed rS318525: Fix MPIC mask/unmask.
Fix MPIC mask/unmask
May 19 2017, 8:25 AM
wma closed D10716: Fix MPIC mask/unmask by committing rS318525: Fix MPIC mask/unmask.
May 19 2017, 8:25 AM · ARM
wma closed D10714: Poll PHY status using internal e6000sw registers by committing rS318524: Poll PHY status using internal e6000sw registers.
May 19 2017, 8:24 AM · ARM
wma committed rS318524: Poll PHY status using internal e6000sw registers.
Poll PHY status using internal e6000sw registers
May 19 2017, 8:24 AM
wma committed rS318522: Enable proper configuration of CESA MBUS windows.
Enable proper configuration of CESA MBUS windows
May 19 2017, 8:20 AM
wma closed D10723: Enable proper configuration of CESA MBUS windows by committing rS318522: Enable proper configuration of CESA MBUS windows.
May 19 2017, 8:20 AM · ARM
wma committed rS318521: Improve busy-wait loop during switch phy access in e6000sw.
Improve busy-wait loop during switch phy access in e6000sw
May 19 2017, 8:17 AM
wma closed D10713: Improve busy-wait loop during switch phy access in e6000sw by committing rS318521: Improve busy-wait loop during switch phy access in e6000sw.
May 19 2017, 8:17 AM · ARM

Feb 2 2017

wma committed rS313089: Remove remaining ifdefs from CESA header.
Remove remaining ifdefs from CESA header
Feb 2 2017, 6:08 AM

Jan 26 2017

wma committed rS312812: Add dummy functions for Marvell SoC's not equipped with AHCI.
Add dummy functions for Marvell SoC's not equipped with AHCI
Jan 26 2017, 11:14 AM

Jan 25 2017

wma committed rS312751: Extend AHCI_Q_BIT_STRING after adding new quirk for soft.
Extend AHCI_Q_BIT_STRING after adding new quirk for soft
Jan 25 2017, 12:18 PM
wma committed rS312749: Add support for AHCI on ARMADA38X.
Add support for AHCI on ARMADA38X
Jan 25 2017, 10:35 AM
wma closed D9222: Add support for AHCI on ARMADA38X by committing rS312749: Add support for AHCI on ARMADA38X.
Jan 25 2017, 10:35 AM
wma closed D9221: Enable optional soft reset in AHCI by committing rS312748: Enable optional soft reset in AHCI.
Jan 25 2017, 10:33 AM
wma committed rS312748: Enable optional soft reset in AHCI.
Enable optional soft reset in AHCI
Jan 25 2017, 10:33 AM
wma committed rS312747: Setup decoding windows for ARMADA38X.
Setup decoding windows for ARMADA38X
Jan 25 2017, 10:31 AM
wma closed D9220: Setup decoding windows for ARMADA38X AHCI by committing rS312747: Setup decoding windows for ARMADA38X.
Jan 25 2017, 10:31 AM
wma closed D9219: Fix node detection for MBUS windows configuration by committing rS312746: Fix node detection for MBUS windows configuration.
Jan 25 2017, 10:30 AM
wma committed rS312746: Fix node detection for MBUS windows configuration.
Fix node detection for MBUS windows configuration
Jan 25 2017, 10:30 AM
wma committed rS312745: Expand OpenFirmware API with ofw_bus_node_status_okay.
Expand OpenFirmware API with ofw_bus_node_status_okay
Jan 25 2017, 10:28 AM
wma closed D9218: Expand OpenFirmware API with ofw_bus_node_status_okay method by committing rS312745: Expand OpenFirmware API with ofw_bus_node_status_okay.
Jan 25 2017, 10:28 AM
wma closed D9215: Fix SHA256 usage on older CESA versions by committing rS312744: Fix SHA256 usage on older CESA versions.
Jan 25 2017, 10:26 AM
wma committed rS312744: Fix SHA256 usage on older CESA versions.
Fix SHA256 usage on older CESA versions
Jan 25 2017, 10:26 AM
wma committed rS312743: Use SoC ID - based detection in CESA.
Use SoC ID - based detection in CESA
Jan 25 2017, 10:22 AM
wma closed D9247: Use SoC ID - based detection in CESA by committing rS312743: Use SoC ID - based detection in CESA.
Jan 25 2017, 10:22 AM
wma closed D9248: Add misssing Armada38x ID's in CESA attach by committing rS312740: Add misssing Armada38x ID's in CESA attach.
Jan 25 2017, 6:11 AM
wma committed rS312740: Add misssing Armada38x ID's in CESA attach.
Add misssing Armada38x ID's in CESA attach
Jan 25 2017, 6:11 AM
wma committed rS312739: Introduce armada_thermal driver for Armada family platforms.
Introduce armada_thermal driver for Armada family platforms
Jan 25 2017, 6:08 AM