My test environment was screwed up during the last version of the
patchset. I was using a kernel which was built previously and didn't pay
attention to commit hash suffix in `uname -r`.
Human mistake, apologize for that.
Alex found it while doing TDX testing on x86/vmware on tip.
Do you want me to resubmit the patchset to do not brake a git bisect?
On 6/25/24 1:47 AM, Borislav Petkov wrote:
On Tue, Jun 25, 2024 at 01:33:48AM -0700, Alexey Makhalov wrote:
Caller of vmware_hypercall_slow() can pass NULL into *out1,
*out2,... *out5. It will lead to a NULL pointer dereference.
Check a pointer for NULL before assigning a value.
I queue your patches and *now* you find this?!
How did you test them in the first place and why was this scenario missed?
Geez.
