Skip to contentSkip to conversation

powerpc/powernv: Make possible for user to force a full ipl cec reboot

PATCH Superseded Vaibhav Jain 3 participants 4 messages Sep 3, 2018 – Sep 5, 2018 PowerPC

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 powerpc/powernv: Make possible for user to force a full ipl cec reboot

Ever since fast reboot is enabled by default in opal, opal_cec_reboot() will use fast-reset instead of full IPL to perform system reboot. This leaves the user with no direct way to force a full IPL reboot except changing an nvram setting that persistently disables fast-reset for all subsequent reboots.

This patch provides a more direct way for the user to force a one-shot full IPL reboot by passing the command line argument 'full' to the reboot command. So the user will be able to tweak the reboot behavior via:

  $ sudo reboot full	# Force a full ipl reboot skipping fast-reset

or

  $ sudo reboot  	# default reboot path (usually fast-reset)

The reboot command passes the un-parsed command argument to the kernel via the 'Reboot' syscall which is then passed on to the arch function pnv_restart(). The patch updates pnv_restart() to handle this cmd-arg and issues opal_cec_reboot2 with OPAL_REBOOT_FULL_IPL to force a full IPL reset.

---
Change-log:

v2 -> Updated the code to handle case when opal_cec_reboot2() is
not supported by opal. [Andrew]

-> Force a call to opal_cec_reboot() is opal_cec_reboot2()
fails or the action verb 'cmd' is not recognized.

---
 arch/powerpc/include/asm/opal-api.h    |  1 +
 arch/powerpc/platforms/powernv/setup.c | 17 ++++++++++++++++-
 2 files changed, 17 insertions(+), 1 deletion(-)
arch/powerpc/include/asm/opal-api.h +1 -0
index 8365353330b4..870fb7b239ea 100644
@@ -1050,6 +1050,7 @@ enum OpalSysCooling {
 enum {
 	OPAL_REBOOT_NORMAL		= 0,
 	OPAL_REBOOT_PLATFORM_ERROR	= 1,
+	OPAL_REBOOT_FULL_IPL		= 2,
 };
 
 /* Argument to OPAL_PCI_TCE_KILL */
arch/powerpc/platforms/powernv/setup.c +16 -1
index adddde023622..650484e0940b 100644
@@ -224,7 +224,22 @@ static void  __noreturn pnv_restart(char *cmd)
 	pnv_prepare_going_down();
 
 	while (rc == OPAL_BUSY || rc == OPAL_BUSY_EVENT) {
-		rc = opal_cec_reboot();
+		/* See if we need to do a full IPL reboot */
+		if (cmd && strcmp(cmd, "full") == 0)
+			rc = opal_cec_reboot2(OPAL_REBOOT_FULL_IPL, NULL);
+		else
+			rc = OPAL_UNSUPPORTED;
+
+		if (!(rc == OPAL_BUSY || rc == OPAL_BUSY_EVENT)) {
+			if (cmd) {
+				pr_warn("Couldn't do a '%s' reboot. Err=%ld\n",
+					cmd, rc);
+				/* reset cmd to force a cec_reboot now */
+				cmd = NULL;
+			}
+			rc = opal_cec_reboot();
+		}
+
 		if (rc == OPAL_BUSY_EVENT)
 			opal_poll_events(NULL);
 		else
Signature · 1 line
2.17.1
Acked-by on v2 1/1
#1 lore raw
Quoted 15 lines · Vaibhav Jain
arch/powerpc/platforms/powernv/setup.c +6 -1
index adddde023622..650484e0940b 100644
@@ -224,7 +224,22 @@ static void  __noreturn pnv_restart(char *cmd)
  	pnv_prepare_going_down();
 
  	while (rc == OPAL_BUSY || rc == OPAL_BUSY_EVENT) {
-		rc = opal_cec_reboot();
+		/* See if we need to do a full IPL reboot */
+		if (cmd && strcmp(cmd, "full") == 0)
+			rc = opal_cec_reboot2(OPAL_REBOOT_FULL_IPL, NULL);
+		else
+			rc = OPAL_UNSUPPORTED;
+

arch/powerpc/platforms/powernv/setup.c static void __noreturn pnv_restart(char *cmd) · line 242

Quoted 10 lines
arch/powerpc/platforms/powernv/setup.c +10 -0
+		if (!(rc == OPAL_BUSY || rc == OPAL_BUSY_EVENT)) {
+			if (cmd) {
+				pr_warn("Couldn't do a '%s' reboot. Err=%ld\n",
+					cmd, rc);
+				/* reset cmd to force a cec_reboot now */
+				cmd = NULL;
+			}
+			rc = opal_cec_reboot();
+		}
+

The logic here is a bit hard to follow, relying on if (cmd && strcmp...) failing then setting rc = OPAL_UNSUPPORTED in order to handle the normal case seems a bit backwards.

I think I would slightly prefer it if the !cmd case were handled first to make it clear that opal_cec_reboot() is the normal case.

But either way:

Signature · 2 lines
Andrew Donnellan              OzLabs, ADL Canberra
[email protected]  IBM Australia Limited
#2 lore raw

Thanks for reviewing this patch Andrew,

Quoted 3 lines · Andrew Donnellan

The logic here is a bit hard to follow, relying on if (cmd && strcmp...) failing then setting rc = OPAL_UNSUPPORTED in order to handle the normal case seems a bit backwards.

Agree. With this change I am trying to make adding new reboot types in future simpler, without resorting to major refactoring of the function. opal_cec_reboot() should be a fallback for any failures in doing a typed reboot. So adding a new foobar reboot type just needs adding a new branch in the if-else ladder.

if (cmd && strcmp(cmd, "full") == 0)

    rc = opal_cec_reboot2(OPAL_REBOOT_FULL_IPL, NULL);
else if(cmd && strcmd(cmd, "foobar") == 0)
    rc = /* new code here */
else
    rc = OPAL_UNSUPPORTED;
Quoted 2 lines

I think I would slightly prefer it if the !cmd case were handled first to make it clear that opal_cec_reboot() is the normal case.

Fair suggestion. But then I will still have to handle the errors just after the if-else ladder is finished to determine if I need to fallback.

Quoted 1 line

Thanks

Signature · 2 lines
Vaibhav Jain <[email protected]>
Linux Technology Center, IBM India Pvt. Ltd.
#3 lore raw
In reply to Vaibhav Jain
Quoted 11 lines · Vaibhav Jain

Ever since fast reboot is enabled by default in opal, opal_cec_reboot() will use fast-reset instead of full IPL to perform system reboot. This leaves the user with no direct way to force a full IPL reboot except changing an nvram setting that persistently disables fast-reset for all subsequent reboots.

This patch provides a more direct way for the user to force a one-shot full IPL reboot by passing the command line argument 'full' to the reboot command. So the user will be able to tweak the reboot behavior via:

.../...

Quoted 7 lines

/* Argument to OPAL_PCI_TCE_KILL */

arch/powerpc/platforms/powernv/setup.c +0 -0
index adddde023622..650484e0940b 100644

arch/powerpc/platforms/powernv/setup.c static void __noreturn pnv_restart(char *cmd) · line 231

Quoted 9 lines
arch/powerpc/platforms/powernv/setup.c +5 -1
@@ -224,7 +224,22 @@ static void  __noreturn pnv_restart(char *cmd)
  	pnv_prepare_going_down();
 
  	while (rc == OPAL_BUSY || rc == OPAL_BUSY_EVENT) {
-		rc = opal_cec_reboot();
+		/* See if we need to do a full IPL reboot */
+		if (cmd && strcmp(cmd, "full") == 0)
+			rc = opal_cec_reboot2(OPAL_REBOOT_FULL_IPL, NULL);
+		else
+			rc = OPAL_UNSUPPORTED;

As discussed offline, please handle OPAL_SUCCESS case well.

-Vasant