protonscr

Steam failed to launch after upgrading Nvidia driver to 580.126.09 version

steamclosed NVIDIA driversDistro Family: DebianWeb Component
ValveSoftware/steam-for-linux#12658 · opened 2026-01-19 by firemeteorxx · updated 2026-05-11 · 13 comments · github
Ffiremeteorxx 2026-01-19 github

Your system information

  • Steam client version (build number or date): Fresh reinstall right before this issue report.
  • Distribution (e.g. Ubuntu): Debian
  • Opted into Steam client beta?: [Yes/No] No
  • Have you checked for system updates?: [Yes/No] No
  • GPU: NVIDIA 3090

Please describe your issue in as much detail as possible:

Just upgraded the NVIDIA driver from 535 series to latest 580.126.09 and found steam no longer working any more. Nuked my ~/.steam directory and started all over again but find no change at all.
It appears to me that webhelper_gpu failed to detect the GPU with the new driver, without any obvious clue to me. But other part of the Steam appears to recognize my card properly:

Running query: 1 - GpuTopology
Response: gpu_topology {
  gpus {
    id: 1
    name: "NVIDIA GeForce RTX 3090"
    vram_size_bytes: 26027753472
    driver_id: k_EGpuDriverId_NvidiaProprietary
    driver_version_major: 580
    driver_version_minor: 126
    driver_version_patch: 9
    luid: 0
  }
  default_gpu_id: 1
}

webhelper-linux.txt
webhelper_gpu.txt

The new driver appears to work just fine otherwise. Only Steam suffers. Any suggestion on how to diagnose further?

Ccyberpunkrocker-zero 2026-01-19 github

I have been wrestling with the same issue with the 580.126.09 driver on Arch Linux.
Steam client segfaults:

steam.sh[19593]: Running Steam on arch rolling 64-bit
steam.sh[19593]: STEAM_RUNTIME is enabled automatically
setup.sh[19645]: Steam runtime environment up-to-date!
steam.sh[19593]: Log already open
steam.sh[19593]: Steam client's requirements are satisfied
CProcessEnvironmentManager is ready, 6 preallocated environment variables.
SaveRegistryVDF
SaveRegistryVDF completed successfully
SaveRegistryVDF
SaveRegistryVDF completed successfully
[2026-01-16 06:59:22] Startup - updater built Dec 23 2025 00:47:59
[2026-01-16 06:59:22] Startup - Steam Client launched with: '/home/madmax/.local/share/Steam/ubuntu12_32/steam' '-srt-logger-opened'
01/16 06:59:22 minidumps folder is set to /tmp/dumps
01/16 06:59:22 Init: Installing breakpad exception handler for appid(steam)/version(1766451605)/tid(19708)
Looks like steam didn't shutdown cleanly, scheduling immediate update check
CProcessEnvironmentManager is ready, 6 preallocated environment variables.
SaveRegistryVDF
SaveRegistryVDF completed successfully
SaveRegistryVDF
SaveRegistryVDF completed successfully
[2026-01-16 06:59:22] Process started with command-line: '/home/madmax/.local/share/Steam/ubuntu12_32/steam' '-child-update-ui' '-child-update-ui-socket' '8' '-srt-logger-opened'
01/16 06:59:22 minidumps folder is set to /tmp/dumps
[2026-01-16 06:59:22] Using update UI: xwin
01/16 06:59:22 Init: Installing breakpad exception handler for appid(steam)/version(0)/tid(19709)
[2026-01-16 06:59:22] Create window
[2026-01-16 06:59:22] Loading cached metrics from disk (/home/madmax/.local/share/Steam/package/steam_client_metrics.bin)
[2026-01-16 06:59:22] Using the following download hosts for Public, Realm steamglobal
[2026-01-16 06:59:22] 1. https://client-update.fastly.steamstatic.com, /, Realm 'steamglobal', weight was 900, source = 'update_hosts_cached.vdf'
[2026-01-16 06:59:22] 2. https://client-update.akamai.steamstatic.com, /, Realm 'steamglobal', weight was 400, source = 'update_hosts_cached.vdf'
[2026-01-16 06:59:22] 3. https://client-update.steamstatic.com, /, Realm 'steamglobal', weight was 1, source = 'baked in'
[2026-01-16 06:59:22] Checking for update on startup
[2026-01-16 06:59:22] Tarkistetaan saatavilla olevat p�ivitykset...
[2026-01-16 06:59:22] Downloading manifest: https://client-update.fastly.steamstatic.com/steam_client_ubuntu12
[2026-01-16 06:59:22] Manifest download: send request
SteamUpdateUI: An X Error occurred
X Error of failed request:  BadValue (integer parameter out of range for operation)
[2026-01-16 06:59:22] Manifest download: waiting for download to finish
[2026-01-16 06:59:22] Manifest download: finished
[2026-01-16 06:59:22] Download skipped: /steam_client_ubuntu12 version 1766451605, installed version 1766451605, existing pending version 0
[2026-01-16 06:59:22] Nothing to do
[2026-01-16 06:59:22] Vahvistetaan asennusta
[2026-01-16 06:59:22] Verifying all executable checksums
[2026-01-16 06:59:23] Verification complete
UpdateUI: skip show logo

Steam logging initialized: directory: /home/madmax/.local/share/Steam/logs

XRRGetOutputInfo Workaround: initialized with override: 0 real: 0xeeeeddc0
XRRGetCrtcInfo Workaround: initialized with override: 0 real: 0xeeeec500
ERROR: unsafe call to unsetenv count:1 var:'DESKTOP_STARTUP_ID'
crash_20260116065924_5.dmp[19717]: Uploading dump (out-of-process)
/tmp/dumps/crash_20260116065924_5.dmp
/home/madmax/.local/share/Steam/steam.sh: line 928: 19708 Segmentation fault         (core dumped) "$STEAMROOT/$STEAMEXEPATH" "$@"
madmax@[neverwinter]:~ $ crash_20260116065924_5.dmp[19717]: Finished uploading minidump (out-of-process): success = yes
crash_20260116065924_5.dmp[19717]: response: CrashID=bp-9b7f3aa9-d553-420a-9c66-674c12260115
crash_20260116065924_5.dmp[19717]: file ''/tmp/dumps/crash_20260116065924_5.dmp'', upload yes: ''CrashID=bp-9b7f3aa9-d553-420a-9c66-674c12260115''

and I'm getting the following in the journal:

ammi 16 06:14:17 neverwinter steam-runtime-steam-remote[15041]: steam-runtime-steam-remote: Steam is not running: No such device or address
tammi 16 06:14:18 neverwinter kernel: NVRM: API mismatch: the client 'steam' (pid 15150)
tammi 16 06:14:20 neverwinter kernel: NVRM: API mismatch: the client 'steam' (pid 15149)
tammi 16 06:14:20 neverwinter kernel: steam[15149]: segfault at 0 ip 00000000ee872d21 sp 00000000ff94356c error 4 in libc.so.6[c3d21,ee7ce000+192000] likely on CPU 1 (core 1, socket 0)
tammi 16 06:14:20 neverwinter systemd-coredump[15161]: Process 15149 (steam) of user 1000 terminated abnormally with signal 11/SEGV, processing...
tammi 16 06:14:20 neverwinter systemd-coredump[15163]: Process 15149 (steam) of user 1000 terminated abnormally without generating a coredump.

Downgrading everything Nvidia-related to previous version 580.119.02 allows Steam to work normally again. And as @firemeteorxx mentioned, only Steam was affected, everything else worked normally.

Kkisak-valve maintainer 2026-01-19 github

In general, the NVIDIA proprietary driver requires that the running kernel module, 32 bit userspace libraries, and 64 bit userspace libraries are all installed and exactly the same version to work properly. kernel: NVRM: API mismatch: the client 'steam' reads like a version mismatch.

Ffiremeteorxx 2026-01-20 github

It appears that my issue differs from @cyberpunkrocker-zero 's, as I didn't see any kernel log about API mismatch.

In my case, the process that keeps crashing is the steamwebhelper. Inside the crashing process, there is only one thread killed with TRAP on a meaningful stack while all others killed with SEGV on a weird stack:

Jan 20 23:46:10 Hasim systemd[1]: Started [email protected] - Process Core Dump (PID 34610/UID 0).
Jan 20 23:46:10 Hasim systemd-coredump[34611]: Removed old coredump core.steamwebhelper.1000.8893e5e7c8f34b5683af36cfdd3f9b39.30387.1768836296000000.zst.
Jan 20 23:46:10 Hasim systemd-coredump[34611]: [🡕] Process 34608 (steamwebhelper) of user 1000 dumped core.
                                               
                                               Stack trace of thread 34608:
                                               #0  0x0000000000000000 n/a (n/a + 0x0)
                                               ELF object binary architecture: AMD x86-64
Jan 20 23:46:10 Hasim systemd[1]: [email protected]: Deactivated successfully.

It appears to me that those crashing is simply a result of the failure of GPU detection in webhelper_gpu. Still don't have clue how to invest further. But let me try to roll back the version a little bit as @cyberpunkrocker-zero did and see what happens.

Ffiremeteorxx 2026-01-20 github

Unfortunately, rolling back to 580.119.02 doesn't make a difference for me.
Also switching to a Debian stock kernel doesn't help. Even though the stock kernel is a bit dated (6.12), this at least rules out the impact from my customized kernel build.

I think I'll need some suggestions to diagnose further. Is there environment variables that can generate more verbose error log from webhelper_gpu?

Really don't want to go all the way back to the old driver version I used before. I was stuck on 535 because of an virgl + qemu related visual effect issue. 535 was the last known good until I find 580 fixed the issue...

Ddablenparty 2026-01-31 github

What about 580.105.08? People have reported that 580.119.02 experiences similar symptoms to the latest 590 driver and I have also personally experienced that. 580.105.08 fixed a lot of issues I had and every time I try to update to a new version, something breaks, and I keep finding myself back on that same driver version.

Ffiremeteorxx 2026-02-01 github

Just tried 105.08 but see no difference, unfortunately. The syndrome appears to be the same -- webhelper_gpu.txt does not recognize the GPU device (VENDOR / DEVICe being 0x0000) in the gpu_info report.

Ffiremeteorxx 2026-02-01 github

I did more experiments with the following versions:

  • 590.48.01
  • 580.126.09
  • 570.211.01
  • 535.288.01

Only the last one works with steam on my machine. All others suffer from the same syndrome as webhelper_gpu process failed to recognize the card.
How do we debug the webhelper_gpu process further?

Gggunterberg 2026-02-10 github

I'm on a gtx 1050 ti, after downgrading from 580.126.09 to 580.105.08 steam now works fine. On 580.126.09 it would simply silently fail.

Steam was working fine on 580.119.02 aswell but I was having other issues with kde on that version so I downgraded further. I'm on Arch Linux.

Edit: tried with aur package on version 580.126.09-2 and steam now opens.

11qh 2026-02-28 github

Fix: sudo apt install nvidia-vaapi-driver (or libva-nvidia-driver on Fedora/Arch)

Root cause: libva calls a NULL function pointer during vaInitialize() when no VA-API backend driver is installed. Chrome's crashpad catches the SIGSEGV and kills the process.

Details and LD_PRELOAD fallback: https://github.com/1qh/steam-troubleshoot

Ffiremeteorxx 2026-03-06 github

Fix: sudo apt install nvidia-vaapi-driver (or libva-nvidia-driver on Fedora/Arch)

Root cause: libva calls a NULL function pointer during vaInitialize() when no VA-API backend driver is installed. Chrome's crashpad catches the SIGSEGV and kills the process.

Details and LD_PRELOAD fallback: https://github.com/1qh/steam-troubleshoot

Thanks for the information. Let me give this a try.

BTW, why does the Kernel version matter if the issue appears to be something wrong in the user land?

Ffiremeteorxx 2026-03-06 github

Unfortunately, installing nvidia-vaapi-driver does not help in my cases.

I confirmed that vainfo gives proper output with the nvidia-vaapi-driver package installed. But the syndrome on Steam does not change.

@1qh , did you see anything fishy in your 'webhelper_gpu.txt' log file? In my case I see report on unrecognized GPU and the process appears to terminate itself with a SIGTRAP instead of SIGSEGV according to the core-dump...

Ffiremeteorxx 2026-03-06 github

I also tried @1qh's LD_PRELOAD hack and didn't get steam back to work.

I can see a lot of extra logs related to the LD_PRELOAD, some reverting the signal-handlers others about 32bit vs 64bit mismatching. Not sure if those are relevant or not but anyway it does not fix my problem...

BTW, I'm on a kernel version that is older than 6.13, in case this matters to @1qh's finding...

(steam-runtime-identify-library-abi:11900): steam-runtime-tools-WARNING **: 23:35:55.081: Reset signal 4 (Illegal instruction) from handler 0x7feac9e32290 to SIG_DFL

(steam-runtime-identify-library-abi:11900): steam-runtime-tools-WARNING **: 23:35:55.081: Reset signal 5 (Trace/breakpoint trap) from handler 0x7feac9e32290 to SIG_DFL

(steam-runtime-identify-library-abi:11900): steam-runtime-tools-WARNING **: 23:35:55.081: Reset signal 11 (Segmentation fault) from handler 0x7feac9e32350 to SIG_DFL
srt-logger[11902]: W: Reset signal 4 (Illegal instruction) from handler 0x7febea548290 to SIG_DFL
srt-logger[11902]: W: Reset signal 5 (Trace/breakpoint trap) from handler 0x7febea548290 to SIG_DFL
srt-logger[11902]: W: Reset signal 11 (Segmentation fault) from handler 0x7febea548350 to SIG_DFL
srt-logger[11903]: W: Reset signal 4 (Illegal instruction) from handler 0x7fae8891f290 to SIG_DFL
srt-logger[11903]: W: Reset signal 5 (Trace/breakpoint trap) from handler 0x7fae8891f290 to SIG_DFL
srt-logger[11903]: W: Reset signal 11 (Segmentation fault) from handler 0x7fae8891f350 to SIG_DFL
srt-logger[11908]: W: Reset signal 4 (Illegal instruction) from handler 0x7f14deb32290 to SIG_DFL
srt-logger[11908]: W: Reset signal 5 (Trace/breakpoint trap) from handler 0x7f14deb32290 to SIG_DFL
srt-logger[11908]: W: Reset signal 11 (Segmentation fault) from handler 0x7f14deb32350 to SIG_DFL
steam-runtime-check-requirements[11938]: W: Reset signal 4 (Illegal instruction) from handler 0x7f759e2e1290 to SIG_DFL
steam-runtime-check-requirements[11938]: W: Reset signal 5 (Trace/breakpoint trap) from handler 0x7f759e2e1290 to SIG_DFL
steam-runtime-check-requirements[11938]: W: Reset signal 11 (Segmentation fault) from handler 0x7f759e2e1350 to SIG_DFL
steam.sh[11829]: Steam client's requirements are satisfied

ERROR: ld.so: object '/nas/src/projects/steam-troubleshoot/steam_cef_gpu_fix.so' from LD_PRELOAD cannot be preloaded (wrong ELF class: ELFCLASS64): ignored.
Ffiremeteorxx 2026-05-11 github

Glad to report that the issue goes away with Nvidia driver version 595.71.05.

Nothing extracted yet.