On 04/03/25 14:51, Harsh Prateek Bora wrote:
On 2/17/25 12:47, Aditya Gupta wrote:
According to PAPR:
R1–7.3.30–3. When the platform receives an ibm,os-term RTAS
call, or
on a system reset without an ibm,nmi-interlock RTAS call, if the
platform has a dump structure registered through the
ibm,configure-kernel-dump call, the platform must process each
registered kernel dump section as required and, when available,
present the dump structure information to the operating system
through the “ibm,kernel-dump” property, updated with status for
each
dump section, until the dump has been invalidated through the
ibm,configure-kernel-dump RTAS call.
If Fadump has been registered, trigger an Fadump boot (memory preserving
boot), if QEMU recieves a 'ibm,os-term' rtas call.
Implementing the fadump boot as:
* pause all vcpus (will save registers later)
* preserve memory regions specified by fadump
Although mentioned later, but needs to call out here as not implemented
in this patch. Ideally, all the prep work patches should be introduced
earlier before enabling the trigger.
Got it, will try rearranging the code. Though with current code, the
trigger won't be called as fadump will not get registered (as of this
patch the rtas call was not exposed to the kernel, this will likely
change in v2).
* do a memory preserving reboot (GUEST_RESET in QEMU doesn't clear
the memory)
Memory regions registered by fadump will be handled in a later patch.
Signed-off-by: Aditya Gupta <adit...@linux.ibm.com>
---
hw/ppc/spapr_rtas.c | 42 ++++++++++++++++++++++++++++++++++++++++++
1 file changed, 42 insertions(+)
diff --git a/hw/ppc/spapr_rtas.c b/hw/ppc/spapr_rtas.c
index eebdf13b1552..01c82375f03d 100644
--- a/hw/ppc/spapr_rtas.c
+++ b/hw/ppc/spapr_rtas.c
@@ -342,6 +342,43 @@ static void
rtas_ibm_set_system_parameter(PowerPCCPU *cpu,
}
struct fadump_metadata fadump_metadata;
+bool is_next_boot_fadump;
+
+static void trigger_fadump_boot(target_ulong spapr_retcode)
+{
+ /*
+ * In PowerNV, SBE stops all clocks for cores, do similar to it
+ * QEMU's nearest equivalent is 'pause_all_vcpus'
+ * See 'stopClocksS0' in SBE source code for more info on SBE part
+ */
+ pause_all_vcpus();
+
+ if (true /* TODO: Preserve memory registered for fadump */) {
+ /* Failed to preserve the registered memory regions */
Instead of this, it is better to introduce the dummy stub here now
which can be populated in a later patch. That also helps in avoiding
code changes in this hunk in future patch.
For eg:
static bool fadump_preserved_mem(void)
{
return false; /* TBD */
}
...
if (!fadump_preserve_mem()) {
...
}
Thanks, makes sense. I will do it this way.
Thanks,
- Aditya Gupta
+ rtas_st(spapr_retcode, 0, RTAS_OUT_HW_ERROR);
+
+ /* Cause a reboot */
+ qemu_system_guest_panicked(NULL);
+ return;
+ }
+
+ /* Mark next boot as fadump boot */
+ is_next_boot_fadump = true;
+
+ /* Reset fadump_registered for next boot */
+ fadump_metadata.fadump_registered = false;
+ fadump_metadata.fadump_dump_active = true;
+
+ /* Then do a guest reset */
+ /*
+ * Requirement:
+ * This guest reset should not clear the memory (which is
+ * the case when this is merged)
+ */
+ qemu_system_reset_request(SHUTDOWN_CAUSE_GUEST_RESET);
+
+ rtas_st(spapr_retcode, 0, RTAS_OUT_SUCCESS);
+}
/* Papr Section 7.4.9 ibm,configure-kernel-dump RTAS call */
static __attribute((unused)) void
rtas_configure_kernel_dump(PowerPCCPU *cpu,
@@ -449,6 +486,11 @@ static void rtas_ibm_os_term(PowerPCCPU *cpu,
target_ulong msgaddr = rtas_ld(args, 0);
char msg[512];
+ if (fadump_metadata.fadump_registered) {
+ /* If fadump boot works, control won't come back here */
+ return trigger_fadump_boot(rets);
+ }
+
cpu_physical_memory_read(msgaddr, msg, sizeof(msg) - 1);
msg[sizeof(msg) - 1] = 0;
