On Thu, Feb 13, 2025 at 11:03 AM Leo Yan [off-list ref] wrote:
On Sun, Feb 02, 2025 at 06:43:04PM -0600, Rob Herring (Arm) wrote:
[...]
quoted
+static void __debug_save_brbe(u64 *brbcr_el1)
+{
+ *brbcr_el1 = 0;
+
+ /* Check if the BRBE is enabled */
+ if (!(read_sysreg_el1(SYS_BRBCR) & (BRBCR_ELx_E0BRE | BRBCR_ELx_ExBRE)))
+ return;
+
+ /*
+ * Prohibit branch record generation while we are in guest.
+ * Since access to BRBCR_EL1 is trapped, the guest can't
+ * modify the filtering set by the host.
+ */
+ *brbcr_el1 = read_sysreg_el1(SYS_BRBCR);
+ write_sysreg_el1(0, SYS_BRBCR);
+}
Should flush branch record and use isb() before exit host kernel?
I don't think so. The isb()'s in the other cases appear to be related
to ordering WRT memory buffers. BRBE is just registers. I would assume
that there's some barrier before we switch to the guest.
I see inconsistence between the function above and BRBE's disable
function. Here it clears E0BRE / ExBRE bits for disabling BRBE, but the
BRBE driver sets the PAUSED bit in BRBFCR_EL1 for disabling BRBE.
Indeed. This works, but the enabled check won't work. I'm going to add
clearing BRBCR to brbe_disable(), and this part will stay the same.
Rob