linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Noam Camus <noamca@mellanox.com>
To: linux-snps-arc@lists.infradead.org
Cc: linux-kernel@vger.kernel.org, Noam Camus <noamca@mellanox.com>
Subject: [PATCH v3 07/11] ARC: [plat-eznps] new command line argument for HW scheduler at MTM
Date: Thu, 15 Jun 2017 11:43:57 +0300	[thread overview]
Message-ID: <1497516241-16446-8-git-send-email-noamca@mellanox.com> (raw)
In-Reply-To: <1497516241-16446-1-git-send-email-noamca@mellanox.com>

From: Noam Camus <noamca@mellanox.com>

We add ability for all cores at NPS SoC to control the number of cycles
HW thread can execute before it is replace with another eligible
HW thread within the same core. The replacement is done by the
HW scheduler.

Signed-off-by: Noam Camus <noamca@mellanox.com>
---
 Documentation/admin-guide/kernel-parameters.txt |    9 ++++
 arch/arc/plat-eznps/mtm.c                       |   46 ++++++++++++++++++++++-
 2 files changed, 53 insertions(+), 2 deletions(-)

diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt
index 15f79c2..5b551f7 100644
--- a/Documentation/admin-guide/kernel-parameters.txt
+++ b/Documentation/admin-guide/kernel-parameters.txt
@@ -2693,6 +2693,15 @@
 			If the dependencies are under your control, you can
 			turn on cpu0_hotplug.
 
+	nps_mtm_hs_ctr= [KNL,ARC]
+			This parameter sets the maximum duration, in
+			cycles, each HW thread of the CTOP can run
+			without interruptions, before HW switches it.
+			The actual maximum duration is 16 times this
+			parameter's value.
+			Format: integer between 1 and 255
+			Default: 255
+
 	nptcg=		[IA-64] Override max number of concurrent global TLB
 			purges which is reported from either PAL_VM_SUMMARY or
 			SAL PALO.
diff --git a/arch/arc/plat-eznps/mtm.c b/arch/arc/plat-eznps/mtm.c
index dcbf8f6..9c78ad6 100644
--- a/arch/arc/plat-eznps/mtm.c
+++ b/arch/arc/plat-eznps/mtm.c
@@ -21,10 +21,13 @@
 #include <plat/mtm.h>
 #include <plat/smp.h>
 
-#define MT_CTRL_HS_CNT		0xFF
+#define MT_HS_CNT_MIN		0x01
+#define MT_HS_CNT_MAX		0xFF
 #define MT_CTRL_ST_CNT		0xF
 #define NPS_NUM_HW_THREADS	0x10
 
+static int mtm_hs_ctr = MT_HS_CNT_MAX;
+
 #ifdef CONFIG_EZNPS_MEM_ERROR_ALIGN
 int do_memory_error(unsigned long address, struct pt_regs *regs)
 {
@@ -127,7 +130,7 @@ void mtm_enable_core(unsigned int cpu)
 	/* Enable HW schedule, stall counter, mtm */
 	mt_ctrl.value = 0;
 	mt_ctrl.hsen = 1;
-	mt_ctrl.hs_cnt = MT_CTRL_HS_CNT;
+	mt_ctrl.hs_cnt = mtm_hs_ctr;
 	mt_ctrl.mten = 1;
 	write_aux_reg(CTOP_AUX_MT_CTRL, mt_ctrl.value);
 
@@ -138,3 +141,42 @@ void mtm_enable_core(unsigned int cpu)
 	 */
 	cpu_relax();
 }
+
+/* Handle an out of bounds mtm hs counter value */
+static void __init handle_mtm_hs_ctr_out_of_bounds_error(uint8_t val)
+{
+	pr_err("** The value must be in range [%d,%d] (inclusive)\n",
+		MT_HS_CNT_MIN, MT_HS_CNT_MAX);
+
+	mtm_hs_ctr = val;
+}
+
+/* Verify and set the value of the mtm hs counter */
+static int __init set_mtm_hs_ctr(char *ctr_str)
+{
+	int ret;
+	long hs_ctr;
+
+	ret = kstrtol(ctr_str, 0, &hs_ctr);
+	if (ret) {
+		pr_err("** Out of range mtm_hs_ctr, using default value %d\n",
+			MT_HS_CNT_MAX);
+		mtm_hs_ctr = MT_HS_CNT_MAX;
+		return -EINVAL;
+	}
+
+	if (hs_ctr > MT_HS_CNT_MAX) {
+		handle_mtm_hs_ctr_out_of_bounds_error(MT_HS_CNT_MAX);
+		return -EDOM;
+	}
+
+	if (hs_ctr < MT_HS_CNT_MIN) {
+		handle_mtm_hs_ctr_out_of_bounds_error(MT_HS_CNT_MIN);
+		return -EDOM;
+	}
+
+	mtm_hs_ctr = hs_ctr;
+
+	return 0;
+}
+early_param("nps_mtm_hs_ctr", set_mtm_hs_ctr);
-- 
1.7.1

  parent reply	other threads:[~2017-06-15  8:45 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-06-15  8:43 [PATCH v3 00/11] plat-eznps upstream cont. set 2 Noam Camus
2017-06-15  8:43 ` [PATCH v3 01/11] ARC: set level of log per CPU during boot to be info level Noam Camus
2017-06-15  8:43 ` [PATCH v3 02/11] ARC: send ipi to all cpus sharing task mm in case of page fault Noam Camus
2017-06-15  8:43 ` [PATCH v3 03/11] ARC: Allow irq threading Noam Camus
2017-08-25 20:45   ` Vineet Gupta
2017-08-25 20:59     ` Thomas Gleixner
2017-08-25 21:02       ` Thomas Gleixner
2017-06-15  8:43 ` [PATCH v3 04/11] ARC: Add CPU topology Noam Camus
2017-06-15  8:43 ` [PATCH v3 05/11] ARC: Support more than one PGDIR for KVADDR Noam Camus
2017-06-15  8:43 ` [PATCH v3 06/11] ARC: [NUMA] added CONFIG_NUMA for plat-eznps Noam Camus
2017-06-15  8:43 ` Noam Camus [this message]
2017-08-21 17:04   ` [PATCH v3 07/11] ARC: [plat-eznps] new command line argument for HW scheduler at MTM Vineet Gupta
2017-08-22  6:16     ` Noam Camus
2017-06-15  8:43 ` [PATCH v3 08/11] ARC: [plat-eznps] Update the init sequence of aux regs per cpu Noam Camus
2017-06-15  8:43 ` [PATCH v3 09/11] ARC: [plat-eznps] Save/Restore extra auxiliary registers Noam Camus
2017-06-15  8:44 ` [PATCH v3 10/11] ARC: [plat-eznps] handle dedicated AUX registers Noam Camus
2017-06-15  8:44 ` [PATCH v3 11/11] ARC: [plat-eznps] avoid toggling of DPC register Noam Camus

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=1497516241-16446-8-git-send-email-noamca@mellanox.com \
    --to=noamca@mellanox.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-snps-arc@lists.infradead.org \
    /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).