linuxppc-dev.lists.ozlabs.org archive mirror
 help / color / mirror / Atom feed
From: Nicholas Piggin <npiggin@gmail.com>
To: linuxppc-dev@lists.ozlabs.org
Cc: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>,
	Nicholas Piggin <npiggin@gmail.com>
Subject: [PATCH] powerpc/powernv: Fix machine check reporting of async store errors
Date: Tue, 18 May 2021 00:03:55 +1000	[thread overview]
Message-ID: <20210517140355.2325406-1-npiggin@gmail.com> (raw)

POWER9 and POWER10 asynchronous machine checks due to stores have their
cause reported in SRR1 but SRR1[42] is set, which in other cases
indicates DSISR cause.

Check for these cases and clear SRR1[42], so the cause matching uses
the i-side (SRR1) table.

Cc: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Fixes: 7b9f71f974 ("powerpc/64s: POWER9 machine check handler")
Fixes: 201220bb0e ("powerpc/powernv: Machine check handler for POWER10")
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---
 arch/powerpc/kernel/mce_power.c | 48 +++++++++++++++++++++++++++------
 1 file changed, 40 insertions(+), 8 deletions(-)

diff --git a/arch/powerpc/kernel/mce_power.c b/arch/powerpc/kernel/mce_power.c
index 667104d4c455..2fff886c549d 100644
--- a/arch/powerpc/kernel/mce_power.c
+++ b/arch/powerpc/kernel/mce_power.c
@@ -481,12 +481,11 @@ static int mce_find_instr_ea_and_phys(struct pt_regs *regs, uint64_t *addr,
 	return -1;
 }
 
-static int mce_handle_ierror(struct pt_regs *regs,
+static int mce_handle_ierror(struct pt_regs *regs, unsigned long srr1,
 		const struct mce_ierror_table table[],
 		struct mce_error_info *mce_err, uint64_t *addr,
 		uint64_t *phys_addr)
 {
-	uint64_t srr1 = regs->msr;
 	int handled = 0;
 	int i;
 
@@ -695,19 +694,19 @@ static long mce_handle_ue_error(struct pt_regs *regs,
 }
 
 static long mce_handle_error(struct pt_regs *regs,
+		unsigned long srr1,
 		const struct mce_derror_table dtable[],
 		const struct mce_ierror_table itable[])
 {
 	struct mce_error_info mce_err = { 0 };
 	uint64_t addr, phys_addr = ULONG_MAX;
-	uint64_t srr1 = regs->msr;
 	long handled;
 
 	if (SRR1_MC_LOADSTORE(srr1))
 		handled = mce_handle_derror(regs, dtable, &mce_err, &addr,
 				&phys_addr);
 	else
-		handled = mce_handle_ierror(regs, itable, &mce_err, &addr,
+		handled = mce_handle_ierror(regs, srr1, itable, &mce_err, &addr,
 				&phys_addr);
 
 	if (!handled && mce_err.error_type == MCE_ERROR_TYPE_UE)
@@ -723,16 +722,20 @@ long __machine_check_early_realmode_p7(struct pt_regs *regs)
 	/* P7 DD1 leaves top bits of DSISR undefined */
 	regs->dsisr &= 0x0000ffff;
 
-	return mce_handle_error(regs, mce_p7_derror_table, mce_p7_ierror_table);
+	return mce_handle_error(regs, regs->msr,
+			mce_p7_derror_table, mce_p7_ierror_table);
 }
 
 long __machine_check_early_realmode_p8(struct pt_regs *regs)
 {
-	return mce_handle_error(regs, mce_p8_derror_table, mce_p8_ierror_table);
+	return mce_handle_error(regs, regs->msr,
+			mce_p8_derror_table, mce_p8_ierror_table);
 }
 
 long __machine_check_early_realmode_p9(struct pt_regs *regs)
 {
+	unsigned long srr1 = regs->msr;
+
 	/*
 	 * On POWER9 DD2.1 and below, it's possible to get a machine check
 	 * caused by a paste instruction where only DSISR bit 25 is set. This
@@ -746,10 +749,39 @@ long __machine_check_early_realmode_p9(struct pt_regs *regs)
 	if (SRR1_MC_LOADSTORE(regs->msr) && regs->dsisr == 0x02000000)
 		return 1;
 
-	return mce_handle_error(regs, mce_p9_derror_table, mce_p9_ierror_table);
+	/*
+	 * Async machine check due to bad real address from store or foreign
+	 * link time out comes with the load/store bit (PPC bit 42) set in
+	 * SRR1, but the cause comes in SRR1 not DSISR. Clear bit 42 so we're
+	 * directed to the ierror table so it will find the cause (which
+	 * describes it correctly as a store error).
+	 */
+	if (SRR1_MC_LOADSTORE(srr1) &&
+			((srr1 & 0x081c0000) == 0x08140000 ||
+			 (srr1 & 0x081c0000) == 0x08180000)) {
+		srr1 &= ~PPC_BIT(42);
+	}
+
+	return mce_handle_error(regs, srr1,
+			mce_p9_derror_table, mce_p9_ierror_table);
 }
 
 long __machine_check_early_realmode_p10(struct pt_regs *regs)
 {
-	return mce_handle_error(regs, mce_p10_derror_table, mce_p10_ierror_table);
+	unsigned long srr1 = regs->msr;
+
+	/*
+	 * Async machine check due to bad real address from store comes with
+	 * the load/store bit (PPC bit 42) set in SRR1, but the cause comes in
+	 * SRR1 not DSISR. Clear bit 42 so we're directed to the ierror table
+	 * so it will find the cause (which describes it correctly as a store
+	 * error).
+	 */
+	if (SRR1_MC_LOADSTORE(srr1) &&
+			(srr1 & 0x081c0000) == 0x08140000) {
+		srr1 &= ~PPC_BIT(42);
+	}
+
+	return mce_handle_error(regs, srr1,
+			mce_p10_derror_table, mce_p10_ierror_table);
 }
-- 
2.23.0


             reply	other threads:[~2021-05-17 14:04 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-05-17 14:03 Nicholas Piggin [this message]
2021-06-24 14:03 ` [PATCH] powerpc/powernv: Fix machine check reporting of async store errors Michael Ellerman

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20210517140355.2325406-1-npiggin@gmail.com \
    --to=npiggin@gmail.com \
    --cc=linuxppc-dev@lists.ozlabs.org \
    --cc=mahesh@linux.vnet.ibm.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).