ChanServ changed the topic of #linux-rockchip to: Rockchip development discussion | Wiki at | Logs at | ML at
ohmic has joined #linux-rockchip
ohmic is now known as _ohm
cnxsoft has joined #linux-rockchip
premoboss has joined #linux-rockchip
indy has quit [Ping timeout: 258 seconds]
indy has joined #linux-rockchip
premoboss has quit [Ping timeout: 248 seconds]
_ohm has left #linux-rockchip ["Leaving"]
JohnDoe_71Rus has joined #linux-rockchip
<ayaka> phh, ping
JohnDoe_71Rus has quit [Quit: KVIrc 4.9.2 Aria]
premoboss has joined #linux-rockchip
<phh> ayaka pong
<ayaka> phh, I received your mail about hevc yesterday
<ayaka> first of all that driver would be merged into mainline, it is not the v4l2
<ayaka> so those problems you ask are not a deal
<phh> ayaka: it doesn't need to go mainline. my aim is to do a dkms
beeno has joined #linux-rockchip
beeno has quit [Client Quit]
<phh> ayaka: i totally understand if you don't want to help
<ayaka> I can help but there are some reasons stopping me to accept those idea
<ayaka> just remove legacy code, it won't be allowed
<ayaka> because the it would effect those SoC
<ayaka> even they don't use kernel 4.4
<phh> well i haven't removed any legacy code
<phh> as far as I can tell, I haven't added any dependency, and I haven't changed any behaviour anywhere
<ayaka> the only problem leaves to you is the iommu
<ayaka> it is quite simple, we need a interface to flush the table in iommu
<phh> yup yupl... I just add a function in rockchip-iommu?
<ayaka> but rk3399 meet a issue with that, so I only apply it to rk3288
<ayaka> yes, or you many create a new common interface
<phh> mmm, that's weird. it looked like an errata for rk3288
<phh> so I don't really understand why it wouldn't work on rk3399
<ayaka> not, it is a problem with rk3399
<phh> then is this flush really needed? if rk3399 can work without it
<ayaka> it is a long story
<ayaka> anyway, rk3288 need it
<phh> ok ok
<ayaka> and because the memory management and allocation is not a standard way
<ayaka> making that problem huge
<ayaka> I am debugging the mali driver currently
<ayaka> it seems not work at latest next kernel
<phh> you mean 4.10, or 4.4 with r13?
<ayaka> 4.10 with r9
<phh> ok, fwiw I had 4.9 with r12 working
<phh> I haven't tried 4.9 with r12 yet
<ayaka> phh, you mean you use the r12 kernel driver?
<phh> yes
<phh> I got it from some time ago
<ayaka> phh, ok maybe it is too old for me to still use r9
<ayaka> I should update to r13
<ayaka> as the r13 mali library is not designed for xserver but wayland
<phh> there are three versions no?
<ayaka> I wanted to use a match kernel driver
<ayaka> phh, rockchip only release r9, r13 mali library
<ayaka> the kernel driver is different
<phh> (rockchip did release r12 as well)
<ayaka> ok it is not from us
<phh> it has fbdev x11 and wayland :P
<ayaka> but anyway, my plan is a v4l2 driver
<phh> I know, but as far as I understand, I guess it will be at the very least 6 months before the API changes will be merged.
<phh> vcodec doesn't require kernel modifications, so it is a better candidate for dkms-ing
<phh> but that's me being inpatient
paulk-collins has joined #linux-rockchip
<phh> ayaka: what do you think of this change: add a flag in the iommu's dts region to specify this iommu needs to invalidate tlb on every map/unmap, and do it in rk_iommu_map/rk_iommu_unmap in rockchip-iommu.c?
<phh> actually, exynos' iommu always invalidate the tlb in unmap
<ayaka> phh, you may just do as exynos does
<phh> ayaka: is there an official documentation or errata about this need for invalidation? I guess this will be needed to be merged mainstream
<phh> ("official" being rockchip)
<ayaka> I forget how I find that
<phh> could that be in the TRM?
<paulk-collins> ayaka, do you haver details about the DPTX firmware in rk3399?
<ayaka> no, no document talk about that
<paulk-collins> ayaka, also, do you know if other firmwares are needed by rk3399?
<paulk-collins> ok
<paulk-collins> it's display port link training firmware
<ayaka> paulk-collins, let me check the TRM
<paulk-collins> thanks
<ayaka> paulk-collins, yes, there is a firmware for DPTX
<paulk-collins> ayaka, do you know what processor runs it?
<paulk-collins> like the ISA?
<ayaka> Xtensa processor
<paulk-collins> nice
<paulk-collins> maybe I will try to write a free firmware replacement
<paulk-collins> this way we can avoid blobs with the kernel
<ayaka> paulk-collins, wow, I even doesn't have a idea about that processor
<paulk-collins> I think it does display port link training
<paulk-collins> I will try to understand how it works
<paulk-collins> ayaka, any information about it is welcome :)
_whitelogger has joined #linux-rockchip
JohnDoe_71Rus has joined #linux-rockchip
mrjay has joined #linux-rockchip
mrjay has quit [Client Quit]
JohnDoe_71Rus has quit [Quit: KVIrc 4.9.2 Aria]
JohnDoe_71Rus has joined #linux-rockchip
mrjay has joined #linux-rockchip
<ayaka> I have not have
<ayaka> have not read the documents of rk3399 fully this time
<ayaka> as I am not in charge of it
<ayaka> phh I remember now
<ayaka> it is introduce to solve a performance problem
<phh> ah...
<phh> so i have to look somewhere else for my corrupiotns
<phh> corruptions*
<ayaka> the old implement would do attach and detach each time to refresh iommu
<ayaka> but Alpha told there is a interface to refresh iommu
<phh> oh ok
<phh> I see
<ayaka> but it break the iommu in rk3399
JohnDoe_71Rus has quit [Quit: KVIrc 4.9.2 Aria]
<ayaka> so the buffer map operation become more complicit
<ayaka> then I don't remember whether that refresh operation is necessary or not
<ayaka> it is really be introduce in a hurry
<ayaka> as it is work well and I want to end the support for the android driver as quick as possible
<ayaka> I mean I want to get rid of it
<ayaka> so I don't verified it again
JohnDoe_71Rus has joined #linux-rockchip
JohnDoe_71Rus has quit [Quit: KVIrc 4.9.2 Aria]
JohnDoe_71Rus has joined #linux-rockchip
cnxsoft has quit [Remote host closed the connection]
JohnDoe0 has joined #linux-rockchip
JohnDoe_71Rus has quit [Ping timeout: 258 seconds]
mrjay has quit [Quit: Page closed]
JohnDoe0 has quit [Quit: KVIrc 4.9.2 Aria]
JohnDoe_71Rus has joined #linux-rockchip
mrjay has joined #linux-rockchip
mrjay has quit [Quit: Page closed]
premoboss has quit [Ping timeout: 240 seconds]
paulk-collins has quit [Quit: Leaving]
JohnDoe_71Rus has quit [Quit: KVIrc 4.9.2 Aria]
Aussie_matt has joined #linux-rockchip