You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Issue reported by Ranjani and reproduced by Pierre.
Reproduce steps:
1 installing Ubuntu,
2. don't copy the fw and tplg binaries are not available under /lib/firmware/intel.
or rename firmware and topology files with an "_" prefix
3. install the SOF kernel with ktest
4. Reboot the system
Expected result:
System can boot although audio cannot work
Actual result:
kernel would just hang at the purple screen and the only way out is to reboot to another kernel version.
I think this is another case of #494 @bardliao@plbossart What do you think
These issues are all triggered by failed DSP probe but success PCI probe.
But our PCI probe now have PM ops that directly required DSP probe result.
staticintsof_suspend(structdevice*dev, boolruntime_suspend)
{
structsof_platform_priv*priv=dev_get_drvdata(dev);
structsnd_sof_dev*sdev=dev_get_drvdata(&priv->pdev_pcm->dev);
intret=0;
// Maybe a simple check NULL for sdev here can solve the problem.
@xiulipan@bardliao@mengdonglin
After checking the dmesg, I think they are the similar issue.
I will write a patch to have check. Do you have the patch now?
Activity
xiulipan commentedon Jan 7, 2019
I think this is another case of #494
@bardliao @plbossart What do you think
These issues are all triggered by failed DSP probe but success PCI probe.
But our PCI probe now have PM ops that directly required DSP probe result.
xiulipan commentedon Jan 7, 2019
bardliao commentedon Jan 7, 2019
@xiulipan Yes, I think we can add a simple check there. And we need to add the check in all sof_pci_pm{}
libinyang commentedon Jan 7, 2019
@xiulipan @bardliao @mengdonglin
After checking the dmesg, I think they are the similar issue.
I will write a patch to have check. Do you have the patch now?
libinyang commentedon Jan 7, 2019
confirmed with my patch, it is caused by the sdev being NULL in sof_suspend.
libinyang commentedon Jan 7, 2019
@xiulipan @bardliao @mengdonglin
I have a patch ready, which can boot successfully, suspend/resume successfully.
xiulipan commentedon Jan 7, 2019
@libinyang
I think the DSP is not actually suspend/resume but just power off/on?
libinyang commentedon Jan 7, 2019
@xiulipan
As fw is not loaded successfully, we can't handle FW. Just suspend/resume in kernel space successfully.
plbossart commentedon Feb 8, 2019
fixed
bpf: Fix out-of-bounds read in check_atomic_load/store()