Boss Posted July 5, 2023 Share #1 Posted July 5, 2023 Hello everyone. it's my first post here on the forum, first of all I anticipate that I'm an Italian boy and therefore I apologize for my English. I am enthusiastic owner of a configuration based on Asrock j5005 + DSM 7.1.1 up 6 (DS3622xs+), ARPL. given the success I've had with this configuration I wanted to embark on a project (perhaps meaningless for many). I built my very first Nas almost for fun using the motherboard of a not too recent laptop but with low power consumption. Here are the details: https://icecat.it/it/p/acer/lu.s780b.041/notebook-portatili-AO751h-52Bw-3931956.html, mounts an Atom Z520 CPU (32bit). I was able to install Nanoboot x86 DSM 5.0-4493 DS214Play recently. Everything works correctly, and I managed to install the latest version of Plex (modifying the package by hand) I can also stream videos in 4K (despite the 10/100 Ethernet card). being that DS214Play and DS514Play are still models supported by DSM 7.X I was thinking (with the help or some advice) of modifying the ARPL to be able to boot on the i686 platform and be able to install DS514Play. I've already started on virtual machine (virtualbox) to prepare files according to the precious advice of @wjz304 here the short discussion: https://github.com/wjz304/arpl-i18n/issues/110. during the first step that was indicated to me during the compilation using the files at this link https://archive.synology.com/download/ToolChain/Synology NAS GPL Source/7.1.1-42962 ( linux-3. x.txz evansport) the compilation starts for the ARM ARMADA platform. Does anyone want to help me in this endeavour? clearly it's a hobby project and not out of necessity, but I'd like to go through with it. if I have a positive response, I will make what I have done available. I'm not a programmer, but I tinker with new and old PCs, Linux, AmigaOS, etc. mainly I'm an electronic ... but I have a lot of perseverance thanks in advance, sorry for the long post... PS: I am attaching my "Platforms.h" where, as suggested to me, I replaced DS918+ with DS514Play. platforms.h Quote Link to comment Share on other sites More sharing options...
Boss Posted July 8, 2023 Author Share #2 Posted July 8, 2023 (edited) trying to compile the 3.x evansport kernel I realized that compared to 3.10 bromolow there are some differences that lead to a different behavior during compilation. so I edited some parameters in the Make file in the root of the folder, and in some lines of the Make file in the arch/x86 subfolder, the compilation happens ending with an error which refers to the missing gcc-10.h file. as i guessed debian installed in the VM is 64bit and this causes problems if the kernel is not cross-compiled. now I'm making a VM with debian 32bit. I don't think anyone cares about this, but I'm determined to get to my goal. Edited July 8, 2023 by Boss Quote Link to comment Share on other sites More sharing options...
Boss Posted November 12, 2023 Author Share #3 Posted November 12, 2023 I managed to compile the 3.2.101 kernel (the latest one available for evansport). in the end you had to cross-compile it (as presumably the more experienced already knew). the problem now, after having made some changes, for example: the word_at_a_time.h file does not exist and having modified here and there (also thanks to the AI) rr-lkms I have arrived at this point. basically to compile rr-klms (from what I understand) you would need at least a 3.10.x kernel since several things are not implemented in 3.2. at the same time I realized that I don't think DMS 7.2 will ever be released for DS415Play and DS214Play, so if I succeeded, eventually, there would be nothing to maintain in the future. it would be nice if @pocopico gave a contribution to a crazy person who doesn't give up but I also think that maybe I should give up... any ideas are welcome. Quote Link to comment Share on other sites More sharing options...
wjz304 Posted November 13, 2023 Share #4 Posted November 13, 2023 First of all, the hardware of arm is very complicated, and rom is not universal. Each hardware needs to compile uboot and separate dts separately, So for now, I don't want to get involved in arm 1 Quote Link to comment Share on other sites More sharing options...
Boss Posted November 13, 2023 Author Share #5 Posted November 13, 2023 Maybe I was a little confused in explaining myself. DS415Play is x86, i386 more precisely i686. It's not ARM. that's why I thought porting some launcher was simple Quote Link to comment Share on other sites More sharing options...
Peter Suh Posted November 13, 2023 Share #6 Posted November 13, 2023 (edited) In the case of DS415play, it is Intel Atom CE5335, Evansport platform. It appears to be based on Intel x86. https://kb.synology.com/en-us/DSM/tutorial/What_kind_of_CPU_does_my_NAS_have#x15-series And just because lkm compiles successfully does not solve the problem. Modules for the Evansport platform should be ready. Of course, you can try it with just the vanilla module, but All platforms additionally developed by ARPL are unable to stabilize this module. (purley, broadwellnkv2, broadwellntbap, braswell) Edited November 13, 2023 by Peter Suh 1 Quote Link to comment Share on other sites More sharing options...
Boss Posted November 13, 2023 Author Share #7 Posted November 13, 2023 do you mean that you have little driver coverage for the different platforms (pc)? or do you mean kernel modules for DSM compatibility with ARPL's linux kernel? honestly I haven't decided yet which path to take, I could try compiling on ARPL or choose a TCRP solution... Quote Link to comment Share on other sites More sharing options...
wjz304 Posted November 13, 2023 Share #8 Posted November 13, 2023 Oh, my fault, I mistakenly thought it was an arm, Quote Link to comment Share on other sites More sharing options...
wjz304 Posted November 13, 2023 Share #9 Posted November 13, 2023 You can pull a x86 of the same version_ 64 kernel, Check out the differences between the functions that reported errors, Perhaps there are more errors, 1 Quote Link to comment Share on other sites More sharing options...
Peter Suh Posted November 13, 2023 Share #10 Posted November 13, 2023 1 hour ago, Boss said: do you mean that you have little driver coverage for the different platforms (pc)? or do you mean kernel modules for DSM compatibility with ARPL's linux kernel? honestly I haven't decided yet which path to take, I could try compiling on ARPL or choose a TCRP solution... The platform referred to here does not refer to the type of PC, but to Synology platforms such as apollolake and geminilake. Modules (drivers) compatible with each Synology platform must be compiled separately. Compiled modules should operate stably without kernel panics, but the four platforms mentioned above do not yet cause kernel panics or operate. I blindly tried evansport's lkm compilation with the workflow scripted by wjz304 without adding any models to platforms.h. The conclusion was a failure, but some adjustments may be needed. I'll try my best. https://github.com/PeterSuh-Q3/redpill-lkm/actions/runs/6849287503 1 Quote Link to comment Share on other sites More sharing options...
Peter Suh Posted November 13, 2023 Share #11 Posted November 13, 2023 https://github.com/SirNoish/pkgscripts-ng/blob/master/include/pkg_util.sh#L117 According to your analysis, evansport is divided into i686 family. 1 Quote Link to comment Share on other sites More sharing options...
Boss Posted November 13, 2023 Author Share #12 Posted November 13, 2023 (edited) First of all thank you for even trying then I think it could also be useful on other platforms like SBC, maybe, who knows From what I understand, Evansport is not a simple i386, or rather the Atom CPUs of that type are 1 core 2 threads with a 32bit x86 instruction set as can be seen from the official Sinology toolchain here: https://archive.synology.com/download/ToolChain/toolchain/7.1-42661 Since you've made an attempt, I'll briefly explain my difficulties as a noob. the first difficulty is that when compiling the 3.x kernel (3.2.101) the make file has additional lines compared to the 3.10.x bromolow which point to the default Armada Arm platform (even though it shouldn't). I had to edit the lines where it says cross compile and point to i386 and in the line below the GCC files to cross compile. every time you have to point to the i386 architecture when preparing the modules (as you can also see from the first lines of my terminal screenshot). example: $ export ARCH=i386 $ export CROSS_COMPILE=/home/boss/Scrivania/i686-pc-linux-gnu/bin/i686-pc-linux-gnu- $ make ARCH=i386 modules_prepare if you want I can pass you the compiled kernel without errors, I got as far as compiling the prepared and compiled (not compiled for rr.lkms) by uploading it to mega.nz I also created an updated platform.h file compared to the one at the beginning of the post. the creation of the modules was a problem that I asked myself later, I was trying to follow the schedule that @wjz304 kindly indicated to me on github. Edited November 13, 2023 by Boss Quote Link to comment Share on other sites More sharing options...
Boss Posted November 13, 2023 Author Share #13 Posted November 13, 2023 (edited) 4 hours ago, wjz304 said: You can pull a x86 of the same version_ 64 kernel, Check out the differences between the functions that reported errors, Perhaps there are more errors, I will also try this route. I don't deny that I had already made an attempt, but after a while I got lost among all the files and code. I even thought about disassembling the x86 nanoboot that launches DSM 5.0 to try to make a Frankenstein, but I told myself that it's all stuff that now with DSM 7 it's practically useless to do it, although it works damn well on my motherboard. Edited November 13, 2023 by Boss Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.