Nvidia filed a fourth Xid 109 bug on October 8, a day after its new Linux driver shipped

Nvidia filed a fourth Xid 109 bug on October 8, a day after its new Linux driver shipped

Nvidia filed a fourth internal bug number against Xid 109 crash reports on October 8, a day after its newest Linux driver shipped without an entry for them. "I have filed a bug 6899071 for tracking purpose," a moderator account wrote at 19:00 UTC in the 615 release feedback topic on Nvidia's developer forum, read at 21:09 UTC on October 8.

Nvidia GeForce RTX 5080 Founders Edition graphics card render

Nvidia's newest Linux driver, 615.78.08, shipped on October 7 with six release-note items and no entry for these crashes. Image: Nvidia

Eleven hours earlier, at 07:35 UTC, a second moderator account told a different reporter the opposite kind of news. "We were unable to reproduce the issue during our internal testing. Could you please share a saved-game file from a point where the issue occurs?" That reporter's crash had drawn a tracking bug of its own, 6869685, on October 4.

Four numbers in one topic, and two replies without one

Counting them needs care, because not every Nvidia reply carries a number. Nvidia's moderators have posted six times in the topic since October 4. Four of those posts state a bug: 6756147 and 6869685 on October 4, 6878084 on October 6, and 6899071 on October 8. The other two state none, one asking a reporter for a driver log and one reporting the failed reproduction. So four numbers from six Nvidia posts in five days.

Two of the four come with a claim worth separating from the rest. For 6756147 and 6878084 the moderator wrote that "we do have internal repro", which is Nvidia saying it can make the crash happen on its own machines. For the other two it said only that the bug had been filed.

What the reports actually describe

The crashes are reports from users, not Nvidia's own account of a defect. They share a kernel signature, Xid 109 with the error string CTX SWITCH TIMEOUT, on driver 615.71.09, in Windows games running on Linux through Proton and vkd3d-proton. One reporter logged 11 hangs in a single session, always in the same area of one game, with the device-lost error coming from vkd3d-proton each time.

The topic they sit in is Nvidia's own feedback thread for the 615 branch, opened on September 9 and standing at 148 posts when it was read. The reporter Nvidia addressed in the first of the four bugs came back on October 7 to say his game was still breaking on the new driver.

The October 8 report that drew the fourth number is a crash on loading a save with path tracing enabled, which the reporter says goes away with the environment variable VKD3D_CONFIG=descriptor_heap. That is one person's observation on one machine, not a fix, and Nvidia has published nothing about a cause.

Linux driver 615.78.08 arrived between the October 6 bug and the October 8 one, on October 7. Its release highlights carry six entries, none of them an Xid 109 or a context-switch timeout, and the one Nvidia leads with is a fix for blocky corruption in The Last of Us Part II. Nvidia never said the release would settle these reports.

The Windows side has had no release in the same window: GeForce Game Ready 617.42 of October 6 was still the newest Game Ready and Studio driver in Nvidia's own download lookup at 21:19 UTC on October 8. What the four numbers buy a reporter is a queue position, and nothing in the topic says how long that queue is.

Leave a Comment

Your email address will not be published. Required fields are marked *