LibreOffice Calc v7.3 fails to open {FIXED - 7.3.1.3-3 in repo}

[elvasked]

HP Desktop 285 G3 - AMD Ryzen 5 2400G

Mate DE upgraded using LO Mgr from ver. 7.2.5.2 to ver. 7.3 all working fine including Calc.

Created new calc files + saved calc files as .ods and .xlsx and opened saved files using LO and Caja.

Code: [Select]

[elvasked@localhost ~]$ inxi -Fxs
System:
  Host: localhost.localdomain Kernel: 5.16.5-pclos1 x86_64 bits: 64 
  compiler: gcc v: 11.2.1 Desktop: MATE 1.26.0 Distro: PCLinuxOS 2022 
Machine:
  Type: Desktop System: HP product: HP 285 G3 MT Business PC v: N/A 
  serial: <superuser required> 
  Mobo: HP model: 8434 v: 11 serial: <superuser required> UEFI: AMI v: F.53 
  date: 08/23/2021 
CPU:
  Info: Quad Core model: AMD Ryzen 5 2400G with Radeon Vega Graphics 
  bits: 64 type: MT MCP arch: Zen rev: 0 cache: L2: 2 MiB 
  flags: avx avx2 lm nx pae sse sse2 sse3 sse4_1 sse4_2 sse4a ssse3 svm 
  bogomips: 57492 
  Speed: 3593 MHz min/max: N/A Core speeds (MHz): 1: 3593 2: 3593 3: 3593 
  4: 3593 5: 1440 6: 3593 7: 3593 8: 3593 
Graphics:
  Device-1: AMD Raven Ridge [Radeon Vega Series / Radeon Vega Mobile Series] 
  vendor: Hewlett-Packard driver: amdgpu v: kernel bus-ID: 38:00.0 
  Display: server: PCLinuxOS 1.21.1.3 driver: loaded: amdgpu,v4l 
  resolution: 1360x768~60Hz 
  OpenGL: renderer: AMD Radeon Vega 11 Graphics (RAVEN DRM 3.44.0 
  5.16.5-pclos1 LLVM 12.0.1) 
  v: 4.6 Mesa 21.3.5 direct render: Yes 
Audio:
  Device-1: AMD Raven/Raven2/Fenghuang HDMI/DP Audio vendor: Hewlett-Packard 
  driver: snd_hda_intel v: kernel bus-ID: 38:00.1 
  Device-2: AMD Raven/Raven2/FireFlight/Renoir Audio Processor 
  vendor: Hewlett-Packard driver: snd_pci_acp3x v: kernel bus-ID: 38:00.5 
  Device-3: AMD Family 17h HD Audio vendor: Hewlett-Packard 
  driver: snd_hda_intel v: kernel bus-ID: 38:00.6 
  Device-4: EasyPass Audioengine D1 type: USB 
  driver: hid-generic,snd-usb-audio,usbhid bus-ID: 1-6:3 
  Sound Server-1: ALSA v: k5.16.5-pclos1 running: yes 
  Sound Server-2: sndio v: N/A running: no 
  Sound Server-3: JACK v: 1.9.18 running: no 
  Sound Server-4: PulseAudio v: 15.0 running: yes 
Network:
  Device-1: Realtek RTL8821CE 802.11ac PCIe Wireless Network Adapter 
  vendor: Hewlett-Packard driver: rtw_8821ce v: N/A port: f000 
  bus-ID: 17:00.0 
  IF: wlan0 state: down mac: b0:68:e6:94:5a:bf 
  Device-2: Realtek RTL8111/8168/8411 PCI Express Gigabit Ethernet 
  vendor: Hewlett-Packard driver: r8169 v: kernel port: e000 bus-ID: 2e:00.0 
  IF: eth0 state: up speed: 1000 Mbps duplex: full mac: c4:65:16:36:a6:24 
Bluetooth:
  Device-1: Realtek Realtek Bluetooth 4.2 Adapter type: USB driver: btusb 
  v: 0.8 bus-ID: 1-10:4 
  Report: hciconfig ID: hci0 rfk-id: 1 state: up address: B0:68:E6:94:5A:C0 
  bt-v: 2.1 lmp-v: 4.2 
Drives:
  Local Storage: total: 352.13 GiB used: 81.95 GiB (23.3%) 
  ID-1: /dev/sda vendor: Crucial model: CT250MX500SSD1 size: 232.89 GiB 
  ID-2: /dev/sdb vendor: Intenso model: SSD size: 119.24 GiB 
Partition:
  ID-1: / size: 49.2 GiB used: 11.01 GiB (22.4%) fs: ext4 dev: /dev/sda5 
  ID-2: /home size: 104.6 GiB used: 19.28 GiB (18.4%) fs: ext4 
  dev: /dev/sda7 
Swap:
  ID-1: swap-1 type: partition size: 4 GiB used: 0 KiB (0.0%) dev: /dev/sda6 
Sensors:
  System Temperatures: cpu: 45.0 C mobo: N/A gpu: amdgpu temp: 44.0 C 
  Fan Speeds (RPM): N/A 
Info:
  Processes: 284 Uptime: 6h 50m Memory: 13.64 GiB used: 2.38 GiB (17.4%) 
  Init: SysVinit runlevel: 5 Compilers: gcc: 11.2.1 clang: 12.0.1 
  Packages: N/A note: see --pkg Shell: Bash v: 5.1.16 inxi: 3.3.06 

[MintSpider]

To test a suggestion I...
a) renamed ./config/libreoffice/ to .config/libreoffice-calc-not-working/
b) ran: libreoffice7.3 --writer (which opened fine, thinking it was for the first time on 7.3)
c) noticed that a new .config/libreoffice/ folder was created.
d) ran: libreoffice7.3 --calc [which failed to open]

Oh well, worth the try.

[tbs]

LO Calc crashes on both of my computers.

My desktop (Bios date 05/31/2016) with Nvidia 510:

Code: [Select]

$ su -c 'inxi -SMCGx'
Password: 
System:    Host: schommer Kernel: 5.16.5-pclos1 x86_64 bits: 64 compiler: gcc v: 11.2.1 Desktop: KDE Plasma 5.23.5 
           Distro: PCLinuxOS 2022 
Machine:   Type: Desktop Mobo: ASRock model: 970 Extreme3 R2.0 serial: N/A UEFI: American Megatrends v: P2.60 
           date: 05/31/2016 
CPU:       Info: 6-Core model: AMD Phenom II X6 1055T bits: 64 type: MCP arch: K10 rev: 0 cache: L2: 3 MiB 
           flags: lm nx pae sse sse2 sse3 sse4a svm bogomips: 33526 
           Speed: 1272 MHz min/max: 800/2800 MHz boost: enabled Core speeds (MHz): 1: 1272 2: 1527 3: 2033 4: 2261 
           5: 2658 6: 902 
Graphics:  Device-1: NVIDIA GM107 [GeForce GTX 750] vendor: ZOTAC driver: nvidia v: 510.47.03 bus-ID: 01:00.0 
           Display: x11 server: PCLinuxOS 1.21.1.3 driver: loaded: nvidia,v4l resolution: 1920x1080~60Hz 
           OpenGL: renderer: NVIDIA GeForce GTX 750/PCIe/SSE2 v: 4.6.0 NVIDIA 510.47.03 direct render: Yes 

I checked TDE and Mate on my desktop pc, and LO Calc crashes on both.

[dpascal]

Tested on my HP-Compaq: unlike my other two machines, LO Calc 7.3.0.3 crash.

DE: Mate
Kernel : 5.4.177.
CPU : Intel Core2 Quad Q8400.
GPU : Nvidia GT 710 with nouveau driver.

Code: [Select]

$ inxi -Cx
CPU:
  Info: Quad Core model: Intel Core2 Quad Q8400 bits: 64 type: MCP 
  arch: Penryn rev: A cache: L2: 2 MiB 
  flags: lm nx pae sse sse2 sse3 sse4_1 ssse3 vmx bogomips: 21280 
  Speed: 2660 MHz min/max: N/A Core speeds (MHz): 1: 2660 2: 2660 3: 2660 
  4: 2488 

[JSladek]

Thought (wishful thinking) that there was a fix in the kernel update to 5.16.7, but no go on Calc.

LO Bug site https://bugs.documentfoundation.org/show_bug.cgi?id=147135 is busy with reports and had a possible fix that didn't pan out.

[svx-biker]

If it's just for the calc part I use gnumeric which is also in the repo.

[dm+]

Based on this: https://wiki.documentfoundation.org/ReleaseNotes/7.3#General_improvements_2

Neumaier summation algorithm⁠Wikipedia logo v3.svg was implemented for using modern SIMD CPU instructions like AVX2 tdf#142307 (Dante Doménech)

and this

traps: soffice.bin[11322] trap invalid opcode ip:7f5a4e0558de sp:7ffcb31ca760 error:0 in libsclo.so[7f5a4dd7e000+e06000]

I would say off-hand that scalc won't run on CPUs lacking AVX (Advanced Vector eXtensions) instructions.

Could the people with "newer" computers check their processor capabilities for avx or avx512 and report back?

Code: [Select]

cat /proc/cpuinfo

[Texstar]

avx and avx2 = calc works for me.

[kjpetrie]

I have avx but not avx2. Calc works here.

[w3ie]

I have avx but not avx2. Calc works here.

+1

[kjpetrie]

We have the same processor!

I have added a comment to the Doc Foundation bug report suggesting others check their flags.

The only query I have on this theory is one of the early reporters said it worked on Windows 10 but crashed on Linux, which tends to suggest not a hardware problem (assuming both OSes were on the same machine), but then it does depend on how the kernel handles the exception. Maybe Windows has a more tolerant attitude and just silently ignores it or substitutes a workaround whereas Linux terminates the program. Programmers could argue for ever about this sort of thing and the best way to protect data etc.

[The CrankyZombie]

I have avx but not avx2. Calc works here.

+1

[tbs]

No avx or avx2 here, and LibreCalc doesn't work for me.

[dm+]

We have the same processor!

I have added a comment to the Doc Foundation bug report suggesting others check their flags.

The only query I have on this theory is one of the early reporters said it worked on Windows 10 but crashed on Linux, which tends to suggest not a hardware problem (assuming both OSes were on the same machine), but then it does depend on how the kernel handles the exception. Maybe Windows has a more tolerant attitude and just silently ignores it or substitutes a workaround whereas Linux terminates the program. Programmers could argue for ever about this sort of thing and the best way to protect data etc.

This is great KJP! For others who want to follow up this drama:
https://bugs.documentfoundation.org/show_bug.cgi?id=147135

[tbs]

Inspired by KJP's comment to the bug report, I added my own as well