lib: sbi_ipi: Validate hartids against domain, not HSM state

Commit 37eaca4 ("lib: sbi_ipi: Return error for invalid hartids")
added a weight check against the HSM-interruptible mask, which excludes
harts in STOPPED state. This causes sbi_ipi_send_many() to return
SBI_EINVAL when the kernel sends a remote fence (e.g. TLB shootdown)
whose hmask includes a hart that was stopped by CPU hotplug:

  __sbi_rfence_v02_call: hbase = [0] hmask = [0x3] failed (error [-22])

Validate requested hartids against the domain's assigned hartmask
instead. This still rejects truly invalid or cross-domain hartids, but
allows HSM-STOPPED harts to be silently skipped — which is safe because
a stopped hart will do a full local TLB flush on its next HSM start
before re-entering S-mode.

Fixes: 37eaca4 ("lib: sbi_ipi: Return error for invalid hartids")
Signed-off-by: Chen Pei <cp0613@linux.alibaba.com>

Author: cp0613

lib/sbi/sbi_ipi.c

@@ -128,17 +128,24 @@ int sbi_ipi_send_many(ulong hmask, ulong hbase, u32 event, void *data)
 
 	if (hbase != -1UL) {
 		struct sbi_hartmask tmp_mask = { 0 };
+		struct sbi_hartmask domain_mask;
 		int count = sbi_popcount(hmask);
 
 		for (i = hbase; hmask; i++, hmask >>= 1) {
 			if (hmask & 1UL)
 				sbi_hartmask_set_hartid(i, &tmp_mask);
 		}
 
-		sbi_hartmask_and(&target_mask, &target_mask, &tmp_mask);
+		/* Validate hartids against domain assignment, not HSM state */
+		rc = sbi_domain_get_assigned_hartmask(dom, &domain_mask);
+		if (rc)
+			return rc;
 
-		if (sbi_hartmask_weight(&target_mask) != count)
+		sbi_hartmask_and(&domain_mask, &domain_mask, &tmp_mask);
+		if (sbi_hartmask_weight(&domain_mask) != count)
 			return SBI_EINVAL;
+
+		sbi_hartmask_and(&target_mask, &target_mask, &tmp_mask);
 	}
 
 	/* Send IPIs */