shibby Posted September 22, 2021 Share #1851 Posted September 22, 2021 (edited) check "dmesg" BTW you have Wireless on your motherboard. So maybe DSM already "see" two network cards and atlantic is the 3th (ds918+ has limit to two network card). Then you have to go to /etc/synoinfo.conf and change maxlanport from 2 to 3. Of course you have to also add mac3 to grub.cfg Edited September 22, 2021 by shibby Link to comment Share on other sites More sharing options...
pocopico Posted September 22, 2021 Share #1852 Posted September 22, 2021 2 minutes ago, stefauresi said: 0000:01:00.0 Class 0200: Device 1d6a:07b1 (rev 02) Subsystem: Device 1043:8741 Kernel driver in use: atlantic So driver is loaded for your card. Verify ifconfig reports your interface and check # ifconfig # ethtool ethX Link to comment Share on other sites More sharing options...
stefauresi Posted September 22, 2021 Share #1853 Posted September 22, 2021 il y a 1 minute, pocopico a dit : So driver is loaded for your card. Verify ifconfig reports your interface and check # ifconfig # ethtool ethX only see 2 NIC intel (motherboard) no lights is on , on my 10Gb NIC Link to comment Share on other sites More sharing options...
shibby Posted September 22, 2021 Share #1854 Posted September 22, 2021 1 minute ago, stefauresi said: only see 2 NIC intel (motherboard) did you read my post above? Link to comment Share on other sites More sharing options...
stefauresi Posted September 22, 2021 Share #1855 Posted September 22, 2021 il y a 2 minutes, shibby a dit : did you read my post above? I don't seen message about atlantic Link to comment Share on other sites More sharing options...
shibby Posted September 22, 2021 Share #1856 Posted September 22, 2021 So maybe take a look once again Link to comment Share on other sites More sharing options...
stefauresi Posted September 22, 2021 Share #1857 Posted September 22, 2021 il y a 12 minutes, shibby a dit : So maybe take a look once again reboot DSM = lights it's ok on 10Gb NIC then after enter this command : insmod /usr/lib/modules/atlantic.ko = no lights I think is not a good driver for my 10Gb NIC Link to comment Share on other sites More sharing options...
shibby Posted September 22, 2021 Share #1858 Posted September 22, 2021 i have XG-C100C on 6.2.3 with Juns Loader. I just told you the steps which i had to do to run 10Gbps NIC on it. My motherboard has 2x NIC, so i had to modify /etc/synoinfo.cfg file to allow DS918+ create 3th LAN connection (by default DS918+ has a limit to only 2 NIC). i didn`t update my Xpenology do redpill because i have and issue to support my second SATA controller. So i can`t check is this atlantic.ko module works. Link to comment Share on other sites More sharing options...
Lazy775 Posted September 22, 2021 Share #1859 Posted September 22, 2021 (edited) Hey, just compiled (and tried like, every redist online) RedPill and I'm stuck on 'starting kernel'. This is the full output: Quote Booting RedPill DS918+ [VERSION v6.2.4 to 7.0.1 tested] Loading Linux... Loading initramfs... Starting kernel I get 0 output from serial port (I may have wired them incorrectly because I shoved a male connector to my board (also male)'s serial connector) Is it like, BIOS specific setting in place or smtg? *spec LSI 9280-8i Asrock J4105 Problem persists with and without HDD Edited September 22, 2021 by Lazy775 Markdown not supported Link to comment Share on other sites More sharing options...
Aigor Posted September 22, 2021 Share #1860 Posted September 22, 2021 46 minutes ago, Lazy775 said: Hey, just compiled (and tried like, every redist online) RedPill and I'm stuck on 'starting kernel'. This is the full output: I get 0 output from serial port (I may have wired them incorrectly because I shoved a male connector to my board (also male)'s serial connector) Is it like, BIOS specific setting in place or smtg? *spec LSI 9280-8i Asrock J4105 Problem persists with and without HDD Have you used a null-modem connection? If you want read from serial port thru another serial port you need a serial cable called "null-modem cable" search on google "null modem cable" Link to comment Share on other sites More sharing options...
toyanucci Posted September 22, 2021 Share #1861 Posted September 22, 2021 (edited) 1 hour ago, Lazy775 said: Hey, just compiled (and tried like, every redist online) RedPill and I'm stuck on 'starting kernel'. This is the full output: I get 0 output from serial port (I may have wired them incorrectly because I shoved a male connector to my board (also male)'s serial connector) Is it like, BIOS specific setting in place or smtg? *spec LSI 9280-8i Asrock J4105 Problem persists with and without HDD That's the exact same motherboard I have. They say it takes about 10-15 minutes for boot after you see starting kernel (so maybe you can try waiting 15 minutes or so then check your router if you see a dhcp request for Diskstation), but like you I tried all the different ones I could find and even after 30 minutes I couldn't connect to it. I'm thinking maybe it's the onboard nic that's the problem? 🤷🏾♂️ Edited September 22, 2021 by toyanucci Link to comment Share on other sites More sharing options...
dhlsam Posted September 22, 2021 Share #1862 Posted September 22, 2021 On 9/10/2021 at 1:03 AM, yanjun said: ixgbe.ko.gz 3.31 MB · 50 downloads ixgbevf.ko 84.3 kB · 44 downloads for 918+ 7.0.1 test fine. I cp ixgbevf.ko into /usr/lib/modules/ folder then insmod /usr/lib/modules/ixgbevf.ko insmod: ERROR: could not insert module /usr/lib/modules/ixgbevf.ko: Invalid module format the dmesg [ 989.270284] ixgbevf: version magic '4.4.59+ SMP preempt mod_unload ' should be '4.4.180+ SMP mod_unload ' [ 1004.399371] ixgbevf: version magic '4.4.59+ SMP preempt mod_unload ' should be '4.4.180+ SMP mod_unload ' [ 1032.991768] ixgbevf: version magic '4.4.59+ SMP preempt mod_unload ' should be '4.4.180+ SMP mod_unload ' Link to comment Share on other sites More sharing options...
WiteWulf Posted September 22, 2021 Share #1863 Posted September 22, 2021 1 hour ago, Lazy775 said: Hey, just compiled (and tried like, every redist online) RedPill and I'm stuck on 'starting kernel'. Redpill does not output anything to the screen after this message. All further diagnostic information is output to the serial port. @ThorGroup *please* put a message on there saying there's no more output, like Jun's does 2 Link to comment Share on other sites More sharing options...
WiteWulf Posted September 22, 2021 Share #1864 Posted September 22, 2021 (edited) Little update on the docker kernel panics thing some of us were diagnosing last week: - @Orphée and "labrous" commented on the github issue that they were seeing similar issues with the Moments app on 6.2.4 and Photos on 7.0.1 (lock ups on VMs, kernel panics on baremetal) - the common thing I'm seeing here is databases: most of the docker crashes I've seen have involved mariadb, influxdb, mysqld or elasticsearch Edited September 22, 2021 by WiteWulf 1 Link to comment Share on other sites More sharing options...
Orphée Posted September 22, 2021 Share #1865 Posted September 22, 2021 (edited) 4 minutes ago, WiteWulf said: Little update on the docker kernel panics thing some of us were diagnosing last week: - @Orphée and "labrous" commented on the github issue that they were seeing similar issues with the Moments app on 6.2.4 and Photos on 7.0.1 (lock ups on VMs, kernel panics on baremetal) - the common thing I'm seeing here is databases: all the docker crashes I've seen have involved mariadb, influxdb, mysqld or elasticsearch It happened randomly with a simple nginx container too... (BUG: soft lockup - CPU#1 stuck for 41s! [runc:19580]) And maybe this IPv6 "issue ?" too while starting docker. Edited September 22, 2021 by Orphée 2 Link to comment Share on other sites More sharing options...
pocopico Posted September 22, 2021 Share #1866 Posted September 22, 2021 (edited) @dhlsam. @toyanucciand everyone with nic issues on baremetal. A small suggestion I have to make, is to boot with a live Linux image and see what are the loaded modules for your system before looking to load a module. Modules with wrong magic number will not be loaded. Edited September 22, 2021 by pocopico 1 Link to comment Share on other sites More sharing options...
dhlsam Posted September 22, 2021 Share #1867 Posted September 22, 2021 8 minutes ago, pocopico said: @dhlsam. @toyanucciand everyone with mic issues on baremetal. A small suggestion I have to make, is to boot with a live Linux image and see what are the loaded modules for your system before looking to load a module. Modules with wrong magic number will not be loaded. Thanks,My 10G nic using ixgbe Link to comment Share on other sites More sharing options...
nemesis122 Posted September 22, 2021 Share #1868 Posted September 22, 2021 (edited) Hi i have Asus 87iplus with 6 sata ports at the same controller when all hdd are connected i can install DSM 3615 redpill 7 but when not all HDD are connected i have the error message that 6 sata ports are disabled so i need the line in the grub.cfg sataports= 6 and i dont know exactly in which i have to write this line. could you help me ? menuentry 'RedPill DS3615xs v7.0.1-42214 RC (USB, Verbose)' { savedefault set root=(hd0,msdos1) echo Loading Linux... linux /zImage mac1=9C69Bxxxxxx netif_num=1 earlycon=uart8250,io,0x3f8,115200n8 syno_hdd_powerup_seq=0 vid=0x0951 syno_hdd_detect=0 pid=0x1665 console=ttyS0,115200n8 elevator=elevator sn=14Bxxxxxxxxxxxx root=/dev/md0 earlyprintk loglevel=15 log_buf_len=32M syno_port_thaw=1 HddHotplug=0 withefi syno_hw_version=DS3615xs vender_format_version=2 echo Loading initramfs... initrd /rd.gz echo Starting kernel with USB boot } Edited September 22, 2021 by nemesis122 Link to comment Share on other sites More sharing options...
WiteWulf Posted September 22, 2021 Share #1869 Posted September 22, 2021 BTW, in testing the kernel panic stuff I installed "Photos" to my machine and was pleasantly surprised to find the face detection is working 👍 1 Link to comment Share on other sites More sharing options...
pocopico Posted September 22, 2021 Share #1870 Posted September 22, 2021 (edited) 50 minutes ago, WiteWulf said: Little update on the docker kernel panics thing some of us were diagnosing last week: - @Orphée and "labrous" commented on the github issue that they were seeing similar issues with the Moments app on 6.2.4 and Photos on 7.0.1 (lock ups on VMs, kernel panics on baremetal) - the common thing I'm seeing here is databases: most of the docker crashes I've seen have involved mariadb, influxdb, mysqld or elasticsearch If you enable the indexing on the universal search, a 3615xs on 7.0.1 system will hang as well i guess on synoelasticd. This has been tested and will not happen on 918+ Edited September 22, 2021 by pocopico 1 Link to comment Share on other sites More sharing options...
WiteWulf Posted September 22, 2021 Share #1871 Posted September 22, 2021 (edited) 2 minutes ago, pocopico said: If you enable the indexing on the universal search, a 3615xs on 7.0.1 system will hang as well i guess on synoelasticd Yes, that's exactly what I observed. It was fine with a small folder of ~250 photos, but when I pointed it at a 2TB folder of music files it kernel panic'd after a few minutes. Edited September 22, 2021 by WiteWulf 1 Link to comment Share on other sites More sharing options...
Popular Post ThorGroup Posted September 22, 2021 Author Popular Post Share #1872 Posted September 22, 2021 (edited) Update time! 💊 This week the updates are going to be less exciting but nonetheless very important, and the first point talks about that in more details. THE BUILD HAS CHANGED - READ ME We're putting this as the first and in red so people see it. The LKM build now requires a version to be specified. We know this will break the @haydibe toolchain but we're sure he will fix it quickly as this is a small change. For reasons why read further. Also, if you used a hack to "fix" the Info Center on 3515xs make sure to read "Hardware monitoring emulation" section carefully. A roadmap Lets first address why this week is less exciting: we spent most of the time testing various scenarios and tracking hidden-yet-missed bugs. This is most likely the second to last round of updates before we release a beta version. The beta version can be used in a day-to-day testing and should be free of dangerous crashes (ok, maybe except that strange kernel lookup - we're looking into it). In the beta some features may still be missing but overall we're already testing a few instances and it's looking good We don't have an official roadmap as someone here asked. However, this is the rough list of things we want to address in that order: Custom drivers support*: there's a hack discussed here to add drivers to a ramdisk. However, this isn't a sustainable solution for many reasons. Mainly the reason is a very limited size. The current ramdisk, especially on systems without compression, is almost at the limit and on slower systems can crash with even a few megabytes more added to it. This seems to be especially unstable on v7 Linux v4 kernel. The solution we're working on will simply include modules on the partition which are loaded in pre-boot and copied to the main OS partition on boot. Naturally from the perspective of the user it will involve simply adding a line to user_conf and dropping a file in "custom". For people digging deeper it will be also much easier as this involves no repacking: simply mount the partition and drop more files in and they will be copied 1:1 and loaded as needed. Full SAS support*: while cheap HBA cards are great and something we actually will recommend for home builds (as they're cheap, stable, and offer 4-6 ports), LSI is king for bigger arrays. We want to make sure these are working as intended making both DSM and Linux happy (which isn't easy in v7 ;)). UEFI support*: there are board which cannot boot in legacy mode. We want to fully test this and offer an official UEFI support in the RedPill. Add support for newest revisions of the OS to 918+ and 3615xs* Step-by-step instruction*/**: essentially we want you guys to help us test not only the code but also a guide It will be published alongisde the beta version. We will prepare it in a form of a GH page so that we can easily incorporate feedback (as we cannot edit posts after ~1h). After we resolve problems with it we will go ahead and publish an RC version on the main forum so that it becomes an official version released for usage by less experienced folks who still want to poke around the OS. Integrate build process into the main repository**: @haydibe is doing an AMAZING job here and we cannot thank him enough. As we discussed with him before we want to incorporate the docker directly into the redpill-load repo and enable LKM builds on GH, so that everything is automated and synchronized as soon as we publish an update. Small fixes: add native support for CPU governor, fix CPU detection to match the model (as some things on v7 react strangely if the CPU doesn't match the model), fix various internal code-smells in the kernel code Add support for GPU acceleration: in short we want to explore adding platform like DVA3221 AMD support: let's just say we aren't fans of Intel recently We want to add a model which natively supports AMD CPUs Work on making mfgBIOS LEDs/buzzers/hwmon stuff more real or at least publish an interface which can be used by others to add an arduino or similar. Things marked with * are aimed to be done by the beta release. Things marked with ** we are HOPING to get done before the beta but we cannot promise Fixed I/O scheduler not found errors Many people reported these, and we finally found some time to look at them. Fixing them should fix the disk scheduling and improve performance in some scenarios. All in all you shouldn't ever see them from now on. Commit: https://github.com/RedPill-TTG/redpill-lkm/commit/aaa5847c589b84783471f9a600eeb12fbf2dc31c Hardware monitoring emulation One of the big features boosted by syno in v7 is central monitoring. However, with that it looks like in v7 the hardware monitoring and things around that has changed a lot from v6. In short v6 allowed for things being out of whack here and there with the hardware sensors while v7 assumes all sensors are in order... and when they're not things break unexpectedly. For example that bug with General tab not loading in Info Center was just a tip of an iceberg. It was just a symptom of a MUCH MUCH bigger issue. If you do "cat /run/hwmon/*" on a DSM loaded with Jun's loader you will get: # cat /run/hwmon/* { "CPU_Temperature": { } } This isn't going to fly on v7. The Info Center is just the beginning of problems. High scemd usage at times was also caused by the lack of hwmon emulation (which again, wasn't strictly needed on v6). After painstakingly analyzing the synobios.h from the kernel and toolchains while also scratching our heads at how one of the drivers for ADT chip (fan controller on a real DS) was modified we think we've nailed it: # cat /run/hwmon/* { "CPU_Temperature": { "CPU_0": "61", "CPU_1": "61" } } { "System_Fan_Speed_RPM": { "fan1_rpm": "997", "fan2_rpm": "986" } } { "System_Thermal_Sensor": { "Remote1": "34", "Local": "32", "Remote2": "34" } } { "System_Voltage_Sensor": { "VCC": "11787", "VPP": "1014", "V33": "3418", "V5": "5161", "V12": "11721" } } The output above comes from a 3615xs instance. On 918+ you will see only CPU_Temperature and hdd backplane status as physically 918+ does not monitor anything else. You can read more about that in the new document in the research repo: https://github.com/RedPill-TTG/dsm-research/blob/master/quirks/hwmon.md All values above are fully emulated as on a VM we have no way of reading them and on a real hardware the only standardized one which we can pass along is the CPU temp (as voltages and thermal sensors vary widely between motherboards). We have a "todo" to make CPU temperature real on bare metal. The ugly part, which forced us to change the LKM build to require major version of the OS to be specified, is that syno changed so many things between v6 and v7 that they forgot to make proper headers. As you can see in the https://github.com/RedPill-TTG/dsm-research/blob/master/quirks/mfgbios.md document they just like that moved two fields in the middle of the struct. You never ever do that. Not without at least some ifdef (which they did for other things). The issue is that change we've found is not present in all toolchains, it's not documented anywhere and we discovered it by accident by comparing files. There's really no sane way to detect that (we've tried really hard). This is why we were forced to add (ugly) requirement for v6/v7 to be specified while building the kernel module. If we find a saner way we will go back and make it automatic again. With the hwmon being emulated there are two VERY important things: you CANNOT disable "supportsystemperature" and "supportsystempwarning"! The hack someone posted here with sed - https://gist.github.com/Izumiko/26b8f221af16b99ddad0bdffa90d4329 - should NOT be used. We cannot stress this enough. Anyone who used it and keeps the installation MUST put "supportsystemperature: yes" and "supportsystempwarning: yes" into user_config at least for a single boot. This is because these values (with =no) are now baked into your OS partition and you MUST fix them to the default "yes". As the loader script doesn't have them specified anywhere it will not touch them anymore. you CANNOT disable ADT support. This comes back way to the Jun's loader and deleting "supportadt7490" on 3615xs. If you're not installing from scratch on a new disk but playing with the same on you MUST put "supportadt7490: yes" to your user_config at least for one boot. Commits: https://github.com/RedPill-TTG/redpill-lkm/commit/72cb5c3620711bcabcfe6e1c67ebfbeca6f7e6bf (mfgBIOS structures v6/v7 changes) https://github.com/RedPill-TTG/redpill-lkm/commit/16adab519d6c4cc604e3fb680fb041e8ddd2167e (HWMON definitions) https://github.com/RedPill-TTG/redpill-lkm/commit/7d95ce2ec7d35493227196be58bbd2c7ca2704f8 (HW capabilities emulation) https://github.com/RedPill-TTG/redpill-lkm/commit/014f70e7127abdf28b47fe58f35086296b5d78ca (HWMON emulation) https://github.com/RedPill-TTG/redpill-load/commit/78c7bf9f903a73d51013a859f590c09ee835f8e7 (remove legacy ADT hack) Fix RTC power-on bug On 918+ the scemd complained about auto power-on setting failing. Now this is fixed and scemd no longer triggers an internal error state. This was happening when OS was sending an empty schedule. Commit: https://github.com/RedPill-TTG/redpill-lkm/commit/80a2879aae23d611228c8791b35e237bfa0ec7a8 Add support for emulated SMART log directory This is an internal change in the SMART emulation layer. Syno uses slightly out-of-spec approach (again, hello v7 mess) and assumes all disks implement one of the ATAPI functionalities which is actually optional. This may actually cause some real disks to fail as well on a real v7 so they will probably eventually fix it themselves. For the time being we did it in RP. Commit: https://github.com/RedPill-TTG/redpill-lkm/commit/278ee99a4dcb8fb8f025915118e7df93824e9bfc ------------------------------------------------------------------------ Today's update is rather short as HWMON layer took us a ton of time to properly research and figure out how to actually do it. There were literally 5 different implementation until we got it right. That being said the forum deleted our responses to comments which we were typing here so we will write them again... but probably tomorrow morning. We will try to sit together tomorrow morning (ok, developer's morning ;)) and write responses again. @WiteWulf Can you try deleting the line with "register_pmu_shim" from redpill_main.c (in init_(void) function) and rebuilding the kernel module? You can then use inject_rp_ko.sh (lkm repo) script to inject it into an existing loader image or rebuild the image. With that you shouldn't have PMU emulation anymore (so the instance can be killed in ~24-48h) but we can see if it's kernel panicking. Edited September 22, 2021 by ThorGroup 11 13 Link to comment Share on other sites More sharing options...
WiteWulf Posted September 22, 2021 Share #1873 Posted September 22, 2021 11 minutes ago, ThorGroup said: Can you try deleting the line with "register_pmu_shim" from redpill_main.c (in init_(void) function) and rebuilding the kernel module? You can then use inject_rp_ko.sh (lkm repo) script to inject it into an existing loader image or rebuild the image. With that you shouldn't have PMU emulation anymore (so the instance can be killed in ~24-48h) but we can see if it's kernel panicking. I'm just about to knock off for the day (and it's my partner's birthday, so no hacking tonight! ), but I'll give that a go tomorrow morning (UK time). Link to comment Share on other sites More sharing options...
pocopico Posted September 22, 2021 Share #1874 Posted September 22, 2021 (edited) I have built and tested both DS918 and DS3615 images. - So far the HWMON works as expected and there is no need for the previous synoinfo.conf hack : (supportsystemperature-yes and supportsystempwarning=yes). In case you had previously "supportsystemperature" and "supportsystempwarning" set, rebuilt the loader and edit both /etc.defaults/synoinfo.conf and /etc/synoinfo.conf and set both to yes. - The issue with the soft lockup on synoelasticd remains on DS3615 image (virtual and physical) and will panic the system when you do a universal search. - The issue with "I/O scheduler elevator not found" on mptsas.ko remains @ThorGroup Thanks for your hard work ! The roadmap looks very promissing. Edited September 22, 2021 by pocopico 1 Link to comment Share on other sites More sharing options...
brauner Posted September 22, 2021 Share #1875 Posted September 22, 2021 Where I can download the bootloader for ESXi? Link to comment Share on other sites More sharing options...
Recommended Posts