01:12 <alyssa> robclark: "It turns out.."

01:13 <alyssa> I thought it was pretty public..? =P

01:21 <robclark> alyssa: I guess it depends on if you read git-log or phoronix?

01:26 <alyssa> robclark: I thought Phoronix reads the git-log :D

01:28 <robclark> I think michael has a setup w/ rss feeds.. plus possibly keyword filtering? But maybe not paying attention to email addr other and sender name?

01:32 <robclark> (either that or he never sleeps or has time for anything else)

06:08 <cyrozap> alyssa: Yeah, that's mostly because I've been working on other stuff. More specifically, I've been working on 1) writing file format parsers (BSDL, BXL) in Rust with the "nom" library 2) writing a miasm (https://github.com/cea-sec/miasm) architecture definition for the NDS32 ISA (CPU used in MediaTek WiFi chips and integrated SoC WiFi)

06:11 <cyrozap> The NDS32 arch definition is going to be important for RE-ing MTK WiFi firmware, because miasm can be used as a plugin for Radare and IDA so you get not just the disassembly (which objdump already gives you) but also a navigable block view and (through miasm) symbolic execution.

10:11 <narmstrong> HdkR: Lyude: tomeu: here is the drm-misc-next based tree for Odroid-N2 (including the bifrost MMU patch) https://github.com/superna9999/linux/tree/amlogic/v5.3/odroid-n2-panfrost

10:16 <narmstrong> booting with mainline is painful on the N2, but basically you will need to generate an uImage (mkimage -A arm64 -O linux -T kernel -C none -a 0x1080000 -e 0x1080000 -n linux-next -d arch/arm64/boot/Image uImage) put the uImage and the arch/arm64/boot/dts/amlogic/meson-g12b-odroid-n2.dtb on an sdcard prepared with the N2 u-boot as described in : https://wiki.odroid.com/odroid-n2/software/building_u-boot then you will need

10:16 <narmstrong> to `setenv bootargs console=ttyAML0,115200 root=/dev/mmcblk0p1;fatload mmc 0 0x1080000 uImage;fatload mmc 0 0x1000000 meson-g12b-odroid-n2.dtb;bootm 0x1080000 - 0x1000000` on the u-boot prompt (assuming you have the rootfs on the second partition of the sdcard and the first partition is FAT and contains uImage and meson-g12b-odroid-n2.dtb)

12:07 <shadeslayer> does anyone know where panfrost_ioctl_mmap_bo get's used?

12:07 <shadeslayer> I'm trying to understand how the buffer mapping works

12:07 <shadeslayer> ohhh

12:08 <shadeslayer> it's a macro

12:49 <shadeslayer> does one not need to call drm_gem_free_mmap_offset when closing the GEM handle?

12:50 <shadeslayer> ( assuming it was mapped by userspace )

12:50 <shadeslayer> because I don't see the panfrost drivers or the gem abstraction doing that

12:53 <shadeslayer> etnaviv has etnaviv_gem_free_object which does this, so I'm a bit confused

13:27 <shadeslayer> probably get's done by drm_gem_shmem_free_object I suppose

13:28 <shadeslayer> ah yeah, drm_gem_shmem_free_object -> drm_gem_object_release -> drm_gem_free_mmap_offset

15:03 <alyssa> cyrozap: Neat :o

15:05 <tomeu> on t760 with 64bit, the GPU has trouble already when trying to read from the first byte of the first job descriptor

15:06 <tomeu> so I guess we are missing something in the kernel side when mapping pages for the GPU

15:06 <tomeu> this may be a job for our friends at Arm :)

15:16 <Lyude> narmstrong:yay, thanks! :)

16:25 <bbrezillon> tomeu: just tested your branch, so moving wallpaper job at flush time works \o/

16:25 <bbrezillon> s/moving/creating/

20:59 <alyssa> :tada:

21:11 <anarsoul> ?