x86/amd_node: Fix PCI device reference counting in amd_smn_init()grafted

The local "root" pointer is a temporary variable used during the device
search. Therefore, refcount related to the search iterators should be cleaned
up after the search is complete.

Use the __free() cleanup macro to ensure the refcount is decremented when the
temporary pointer goes out of scope.

Additionally, increment the refcount when caching a root pointer. This ensures
the in-use refcount is separate from the temporary search refcounting.

Finally, drop the redundant "root = NULL" before the second search loop. The
pci_get_class() iterator always decrements the refcount of its "from"
argument, so the first loop can only fall through with "root" already NULL.

Fixes: 0a4b61d9c2e4 ("x86/amd_node: Fix AMD root device caching")
Closes: https://sashiko.dev/#/patchset/20260806160159.230453-1-jason.andryuk%40amd.com
Reported-by: Sashiko <sashiko-bot@kernel.org>
Assisted-by: LLM
Signed-off-by: Yazen Ghannam <yazen.ghannam@amd.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Reviewed-by: Mario Limonciello (AMD) <superm1@kernel.org>
Cc: <stable@kernel.org>
Link: https://patch.msgid.link/20260903154325.74343-1-yazen.ghannam@amd.com

1 files changed, 75 insertions, 0 deletions

diff --git a/arch/xtensa/lib/divsi3.S b/arch/xtensa/lib/divsi3.S
new file mode 100644
index 000000000..edb3c4ad9
--- /dev/null
+++ b/arch/xtensa/lib/divsi3.S
@@ -0,0 +1,75 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later WITH GCC-exception-2.0 */
+#include <linux/linkage.h>
+#include <asm/asmmacro.h>
+#include <asm/core.h>
+
+ENTRY(__divsi3)
+
+	abi_entry_default
+#if XCHAL_HAVE_DIV32
+	quos	a2, a2, a3
+#else
+	xor	a7, a2, a3	/* sign = dividend ^ divisor */
+	do_abs	a6, a2, a4	/* udividend = abs (dividend) */
+	do_abs	a3, a3, a4	/* udivisor = abs (divisor) */
+	bltui	a3, 2, .Lle_one	/* check if udivisor <= 1 */
+	do_nsau	a5, a6, a2, a8	/* udividend_shift = nsau (udividend) */
+	do_nsau	a4, a3, a2, a8	/* udivisor_shift = nsau (udivisor) */
+	bgeu	a5, a4, .Lspecial
+
+	sub	a4, a4, a5	/* count = udivisor_shift - udividend_shift */
+	ssl	a4
+	sll	a3, a3		/* udivisor <<= count */
+	movi	a2, 0		/* quotient = 0 */
+
+	/* test-subtract-and-shift loop; one quotient bit on each iteration */
+#if XCHAL_HAVE_LOOPS
+	loopnez	a4, .Lloopend
+#endif /* XCHAL_HAVE_LOOPS */
+.Lloop:
+	bltu	a6, a3, .Lzerobit
+	sub	a6, a6, a3
+	addi	a2, a2, 1
+.Lzerobit:
+	slli	a2, a2, 1
+	srli	a3, a3, 1
+#if !XCHAL_HAVE_LOOPS
+	addi	a4, a4, -1
+	bnez	a4, .Lloop
+#endif /* !XCHAL_HAVE_LOOPS */
+.Lloopend:
+
+	bltu	a6, a3, .Lreturn
+	addi	a2, a2, 1	/* increment if udividend >= udivisor */
+.Lreturn:
+	neg	a5, a2
+	movltz	a2, a5, a7	/* return (sign < 0) ? -quotient : quotient */
+	abi_ret_default
+
+.Lle_one:
+	beqz	a3, .Lerror
+	neg	a2, a6		/* if udivisor == 1, then return... */
+	movgez	a2, a6, a7	/* (sign < 0) ? -udividend : udividend */
+	abi_ret_default
+
+.Lspecial:
+	bltu	a6, a3, .Lreturn0 /* if dividend < divisor, return 0 */
+	movi	a2, 1
+	movi	a4, -1
+	movltz	a2, a4, a7	/* else return (sign < 0) ? -1 : 1 */
+	abi_ret_default
+
+.Lerror:
+	/* Divide by zero: Use an illegal instruction to force an exception.
+	   The subsequent "DIV0" string can be recognized by the exception
+	   handler to identify the real cause of the exception.  */
+	ill
+	.ascii	"DIV0"
+
+.Lreturn0:
+	movi	a2, 0
+#endif /* XCHAL_HAVE_DIV32 */
+	abi_ret_default
+
+ENDPROC(__divsi3)
+EXPORT_SYMBOL(__divsi3)