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, 139 insertions, 0 deletions

diff --git a/arch/arm/lib/findbit.S b/arch/arm/lib/findbit.S
new file mode 100644
index 000000000..b7ac2d3c0
--- /dev/null
+++ b/arch/arm/lib/findbit.S
@@ -0,0 +1,139 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ *  linux/arch/arm/lib/findbit.S
+ *
+ *  Copyright (C) 1995-2000 Russell King
+ *
+ * 16th March 2001 - John Ripley <jripley@sonicblue.com>
+ *   Fixed so that "size" is an exclusive not an inclusive quantity.
+ *   All users of these functions expect exclusive sizes, and may
+ *   also call with zero size.
+ * Reworked by rmk.
+ */
+#include <linux/linkage.h>
+#include <asm/assembler.h>
+#include <asm/unwind.h>
+                .text
+
+#ifdef __ARMEB__
+#define SWAB_ENDIAN le
+#else
+#define SWAB_ENDIAN be
+#endif
+
+		.macro	find_first, endian, set, name
+ENTRY(_find_first_\name\()bit_\endian)
+	UNWIND(	.fnstart)
+		teq	r1, #0
+		beq	3f
+		mov	r2, #0
+1:		ldr	r3, [r0], #4
+		.ifeq \set
+		mvns	r3, r3			@ invert/test bits
+		.else
+		movs	r3, r3			@ test bits
+		.endif
+		.ifc \endian, SWAB_ENDIAN
+		bne	.L_found_swab
+		.else
+		bne	.L_found		@ found the bit?
+		.endif
+		add	r2, r2, #32		@ next index
+2:		cmp	r2, r1			@ any more?
+		blo	1b
+3:		mov	r0, r1			@ no more bits
+		ret	lr
+	UNWIND(	.fnend)
+ENDPROC(_find_first_\name\()bit_\endian)
+		.endm
+
+		.macro	find_next, endian, set, name
+ENTRY(_find_next_\name\()bit_\endian)
+	UNWIND(	.fnstart)
+		cmp	r2, r1
+		bhs	3b
+		mov	ip, r2, lsr #5		@ word index
+		add	r0, r0, ip, lsl #2
+		ands	ip, r2, #31		@ bit position
+		beq	1b
+		ldr	r3, [r0], #4
+		.ifeq \set
+		mvn	r3, r3			@ invert bits
+		.endif
+		.ifc \endian, SWAB_ENDIAN
+		rev_l	r3, ip
+		.if	.Lrev_l_uses_tmp
+		@ we need to recompute ip because rev_l will have overwritten
+		@ it.
+		and	ip, r2, #31		@ bit position
+		.endif
+		.endif
+		movs	r3, r3, lsr ip		@ shift off unused bits
+		bne	.L_found
+		orr	r2, r2, #31		@ no zero bits
+		add	r2, r2, #1		@ align bit pointer
+		b	2b			@ loop for next bit
+	UNWIND(	.fnend)
+ENDPROC(_find_next_\name\()bit_\endian)
+		.endm
+
+		.macro	find_bit, endian, set, name
+		find_first \endian, \set, \name
+		find_next  \endian, \set, \name
+		.endm
+
+/* _find_first_zero_bit_le and _find_next_zero_bit_le */
+		find_bit le, 0, zero_
+
+/* _find_first_bit_le and _find_next_bit_le */
+		find_bit le, 1
+
+#ifdef __ARMEB__
+
+/* _find_first_zero_bit_be and _find_next_zero_bit_be */
+		find_bit be, 0, zero_
+
+/* _find_first_bit_be and _find_next_bit_be */
+		find_bit be, 1
+
+#endif
+
+/*
+ * One or more bits in the LSB of r3 are assumed to be set.
+ */
+.L_found_swab:
+	UNWIND(	.fnstart)
+		rev_l	r3, ip
+.L_found:
+#if __LINUX_ARM_ARCH__ >= 7
+		rbit	r3, r3			@ reverse bits
+		clz	r3, r3			@ count high zero bits
+		add	r0, r2, r3		@ add offset of first set bit
+#elif __LINUX_ARM_ARCH__ >= 5
+		rsb	r0, r3, #0
+		and	r3, r3, r0		@ mask out lowest bit set
+		clz	r3, r3			@ count high zero bits
+		rsb	r3, r3, #31		@ offset of first set bit
+		add	r0, r2, r3		@ add offset of first set bit
+#else
+		mov	ip, #~0
+		tst	r3, ip, lsr #16		@ test bits 0-15
+		addeq	r2, r2, #16
+		moveq	r3, r3, lsr #16
+		tst	r3, #0x00ff
+		addeq	r2, r2, #8
+		moveq	r3, r3, lsr #8
+		tst	r3, #0x000f
+		addeq	r2, r2, #4
+		moveq	r3, r3, lsr #4
+		tst	r3, #0x0003
+		addeq	r2, r2, #2
+		moveq	r3, r3, lsr #2
+		tst	r3, #0x0001
+		addeq	r2, r2, #1
+		mov	r0, r2
+#endif
+		cmp	r1, r0			@ Clamp to maxbit
+		movlo	r0, r1
+		ret	lr
+	UNWIND(	.fnend)