lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <20241112110522.141924-1-vadfed@meta.com>
Date: Tue, 12 Nov 2024 03:05:22 -0800
From: Vadim Fedorenko <vadfed@...a.com>
To: Vadim Fedorenko <vadim.fedorenko@...ux.dev>,
        Pavan Chebbi
	<pavan.chebbi@...adcom.com>,
        Andrew Lunn <andrew+netdev@...n.ch>, Paolo Abeni
	<pabeni@...hat.com>,
        Michael Chan <michael.chan@...adcom.com>,
        Jakub Kicinski
	<kuba@...nel.org>
CC: Richard Cochran <richardcochran@...il.com>, <netdev@...r.kernel.org>,
        "David S. Miller" <davem@...emloft.net>,
        Eric Dumazet <edumazet@...gle.com>, Vadim Fedorenko <vadfed@...a.com>,
        Simon Horman <horms@...nel.org>
Subject: [PATCH net-next] bnxt_en: optimize gettimex64

Current implementation of gettimex64() makes at least 3 PCIe reads to
get current PHC time. It takes at least 2.2us to get this value back to
userspace. At the same time there is cached value of upper bits of PHC
available for packet timestamps already. This patch reuses cached value
to speed up reading of PHC time.

Signed-off-by: Vadim Fedorenko <vadfed@...a.com>
---
I did some benchmarks on host with Broadcom Thor NIC trying to build
histogram of time spent to call clock_gettime() to query PTP device
over million iterations.
With current implementation the result is (long tail is cut):

2200ns: 902624
2300ns: 87404
2400ns: 4025
2500ns: 1307
2600ns: 581
2700ns: 261
2800ns: 104
2900ns: 36
3000ns: 32
3100ns: 24
3200ns: 16
3300ns: 29
3400ns: 29
3500ns: 23

Optimized version on the very same machine and NIC gives next values:

900ns: 865436
1000ns: 128630
1100ns: 2671
1200ns: 727
1300ns: 397
1400ns: 178
1500ns: 92
1600ns: 16
1700ns: 15
1800ns: 11
1900ns: 6
2000ns: 20
2100ns: 11

That means pct(99) improved from 2300ns to 1000ns.
---
 drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c | 32 +++++++++++++++++--
 1 file changed, 30 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c
index 91e7e08fabb1..8764ce412f7b 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c
@@ -112,6 +112,28 @@ static int bnxt_refclk_read(struct bnxt *bp, struct ptp_system_timestamp *sts,
 	return rc;
 }
 
+static int bnxt_refclk_read_low(struct bnxt *bp, struct ptp_system_timestamp *sts,
+				u32 *low)
+{
+	struct bnxt_ptp_cfg *ptp = bp->ptp_cfg;
+	unsigned long flags;
+
+	/* We have to serialize reg access and FW reset */
+	read_seqlock_excl_irqsave(&ptp->ptp_lock, flags);
+
+	if (test_bit(BNXT_STATE_IN_FW_RESET, &bp->state)) {
+		read_sequnlock_excl_irqrestore(&ptp->ptp_lock, flags);
+		return -EIO;
+	}
+
+	ptp_read_system_prets(sts);
+	*low = readl(bp->bar0 + ptp->refclk_mapped_regs[0]);
+	ptp_read_system_postts(sts);
+
+	read_sequnlock_excl_irqrestore(&ptp->ptp_lock, flags);
+	return 0;
+}
+
 static void bnxt_ptp_get_current_time(struct bnxt *bp)
 {
 	struct bnxt_ptp_cfg *ptp = bp->ptp_cfg;
@@ -162,13 +184,19 @@ static int bnxt_ptp_gettimex(struct ptp_clock_info *ptp_info,
 {
 	struct bnxt_ptp_cfg *ptp = container_of(ptp_info, struct bnxt_ptp_cfg,
 						ptp_info);
-	u64 ns, cycles;
+	u64 ns, cycles, time;
+	u32 low;
 	int rc;
 
-	rc = bnxt_refclk_read(ptp->bp, sts, &cycles);
+	rc = bnxt_refclk_read_low(ptp->bp, sts, &low);
 	if (rc)
 		return rc;
 
+	time = (u64)READ_ONCE(ptp->old_time) << BNXT_HI_TIMER_SHIFT;
+	cycles = (time & BNXT_HI_TIMER_MASK) | low;
+	if (low < (time & BNXT_LO_TIMER_MASK))
+		cycles += BNXT_LO_TIMER_MASK + 1;
+
 	ns = bnxt_timecounter_cyc2time(ptp, cycles);
 	*ts = ns_to_timespec64(ns);
 
-- 
2.43.5


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ