File: datasym.c

package info (click to toggle)
linux 6.1.139-1
  • links: PTS, VCS
  • area: main
  • in suites: bookworm-proposed-updates
  • size: 1,495,880 kB
  • sloc: ansic: 23,469,452; asm: 266,614; sh: 110,522; makefile: 49,887; python: 36,990; perl: 36,834; cpp: 6,056; yacc: 4,908; lex: 2,725; awk: 1,440; ruby: 25; sed: 5
file content (40 lines) | stat: -rw-r--r-- 1,001 bytes parent folder | download | duplicates (11)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
#include <linux/compiler.h>
#include "../tests.h"

typedef struct _buf {
	char data1;
	char reserved[55];
	char data2;
} buf __attribute__((aligned(64)));

static buf buf1 = {
	/* to have this in the data section */
	.reserved[0] = 1,
};

static int datasym(int argc __maybe_unused, const char **argv __maybe_unused)
{
	for (;;) {
		buf1.data1++;
		if (buf1.data1 == 123) {
			/*
			 * Add some 'noise' in the loop to work around errata
			 * 1694299 on Arm N1.
			 *
			 * Bias exists in SPE sampling which can cause the load
			 * and store instructions to be skipped entirely. This
			 * comes and goes randomly depending on the offset the
			 * linker places the datasym loop at in the Perf binary.
			 * With an extra branch in the middle of the loop that
			 * isn't always taken, the instruction stream is no
			 * longer a continuous repeating pattern that interacts
			 * badly with the bias.
			 */
			buf1.data1++;
		}
		buf1.data2 += buf1.data1;
	}
	return 0;
}

DEFINE_WORKLOAD(datasym);