CVE-2026-98313: drm/msm/dp: skip PUSH_IDLE when the link was never enabled
In the Linux kernel, the following vulnerability has been resolved:
drm/msm/dp: skip PUSHIDLE when the link was never enabled
msmdpdisplayatomicenable() returns early when link training fails, leaving ->poweron false and the main link down. msmdpdisplayatomicdisable() nevertheless writes DPSTATECTRLPUSHIDLE and waits for an idle-pattern completion that cannot arrive, so every failed enable is followed by "PUSHIDLE pattern timedout".
Every other step of the teardown is already gated on that flag: msmdpdisplaydisable(), called from .atomicpostdisable(), returns early on !poweron. The PUSHIDLE write is the only one that is not, so the controller's runtime-PM reference is then dropped without the link having been taken down.
On glymur (Snapdragon X2 Elite) the consequence is not a warning. The SoC does not survive it: TrustZone force-stops the SOCCP and ADSP remote processors and the machine resets silently about 50 ms later, with no oops and no panic. On an ASUS Zenbook A16 (UX3607OA), whose eDP panel does not currently train, this reproduces without any compositor or GPU involvement:
# eDP enable has already failed with "Failed link training (rc=-104)" echo 1 > /sys/class/graphics/fb0/blank
[535.645455] === marker === [535.694833] qcomq6v5pas d00000.remoteproc: fatal error received: \ sysmsmsm.c:512:TZ force stop [535.694875] remoteproc remoteproc0: crash detected in soccp: type fatal error [535.728857] qcomq6v5pas 6800000.remoteproc: fatal error received: \ sysmsmsm.c:783:err fatal notification received from TZ <SoC reset>
Gate the PUSHIDLE write on ->poweron so the disable path is consistent with the rest of the teardown. With this applied the same sequence is harmless and the machine stays up; without it, it resets every time.
The unconditional write dates back to the original DP driver (c943b4948b58 ("drm/msm/dp: add displayPort driver support")), but the surrounding code has been restructured several times since, so no Fixes: tag is offered.
Note that the eDP link-training failure that exposes this on the A16 is a separate problem in the glymur eDP PHY and is reported separately; this change is about not damaging the machine when training fails, for whatever reason.
Tested on ASUS Zenbook A16 (UX3607OA), Snapdragon X2 Elite Extreme, on linux-next next-20260803 and next-20260807. The machine has since been running next-20260807 with this patch as its daily driver.
Patchwork: https://patchwork.freedesktop.org/patch/745167/
Affected Software
Remediation
Recommended actions to resolve this vulnerability, in priority order.
- Configuration
Gate the PUSH_IDLE write on ->power_on so the disable path skips it when eDP link training failed and the link was never enabled.
Linux kernel drm/msm/dp DP_STATE_CTRL_PUSH_IDLE write = skip when ->power_on is false
Event History
Frequently Asked Questions
What systems are most likely to be affected?
Systems using the MSM DisplayPort driver are affected when link training fails and the display-enable path exits before the main link is enabled. The reported severe outcome occurs on glymur (Snapdragon X2 Elite); an ASUS Zenbook A16 (UX3607OA) with an eDP panel that does not train also reproduces the failed-enable condition.
Does this require an attacker or compositor/GPU activity to trigger?
No attacker-controlled input or compositor/GPU involvement is described. The issue is triggered when link training has already failed and the subsequent disable path attempts PUSH_IDLE despite the link never having been enabled.
How can I recognize this issue in logs or behavior?
A failed enable may log "Failed link training (rc=-104)", followed by "PUSH_IDLE pattern timedout" during teardown. On the affected Snapdragon X2 Elite system, the result can be a silent machine reset roughly 50 ms later, without an oops or panic.
What should be done if the system is affected?
Use the resolved change referenced by the listed stable kernel commits. The fix skips the PUSH_IDLE operation when the link was never enabled; no separate temporary workaround is provided in the supplied data.