Re: [PATCH] clocksource: Add driver for the Ingenic JZ47xx OST

From: Paul Cercueil
Date: Thu Nov 07 2019 - 14:58:09 EST



Le jeu., nov. 7, 2019 at 20:39, Daniel Lezcano <daniel.lezcano@xxxxxxxxxx> a Ãcrit :
On 07/11/2019 16:56, Paul Cercueil wrote:
Hi Daniel,

On 2019-08-16 16:54, Daniel Lezcano wrote:
On 09/08/2019 14:38, Paul Cercueil wrote:
From: Maarten ter Huurne <maarten@xxxxxxxxxxxxxx>

OST is the OS Timer, a 64-bit timer/counter with buffered reading.

SoCs before the JZ4770 had (if any) a 32-bit OST; the JZ4770 and
JZ4780 have a 64-bit OST.

This driver will register both a clocksource and a sched_clock to the
system.

Signed-off-by: Maarten ter Huurne <maarten@xxxxxxxxxxxxxx>
Signed-off-by: Paul Cercueil <paul@xxxxxxxxxxxxxxx>
Tested-by: Mathieu Malaterre <malat@xxxxxxxxxx>
Tested-by: Artur Rojek <contact@xxxxxxxxxxxxxx>
---


[ ... ]

+ err = clocksource_register_hz(cs, rate);
+ if (err) {
+ dev_err(dev, "clocksource registration failed: %d\n", err);
+ clk_disable_unprepare(ost->clk);
+ return err;
+ }
+
+ /* Cannot register a sched_clock with interrupts on */

Aren't they already disabled?

Sorry for the late reply.

No, they are not already disabled; this is what I get if I comment out
the local_irq_save()/local_irq_restore():

[ 0.361014] clocksource: ingenic-ost: mask: 0xffffffff max_cycles:
0xffffffff, max_idle_ns: 159271703898 ns
[ 0.361515] clocksource: Switched to clocksource ingenic-ost
[ 0.361686] ------------[ cut here ]------------
[ 0.361893] WARNING: CPU: 0 PID: 1 at kernel/time/sched_clock.c:179
sched_clock_register+0x7c/0x2e4
[ 0.362174] CPU: 0 PID: 1 Comm: swapper Not tainted 5.4.0-rc5+ #461
[ 0.362330] Stack : 80744558 80069b44 80770000 00000000 00000000
00dfd7a7 806e6db4 8106bb74
[ 0.362619] 806f0000 81067ca4 806f31c7 80769478 00000020
10000400 8106bb20 00dfd7a7
[ 0.362906] 00000000 00000000 80780000 00000000 00000007
00000001 00000049 3563722d
[ 0.363191] 8106ba61 00000000 ffffffff 00000010 806f0000
00000000 00000000 806f0000
[ 0.363477] 00000020 00000000 80714534 80770000 00000002
80319154 00000000 80770000
[ 0.363762] ...
[ 0.363906] Call Trace:
[ 0.364087] [<8001af14>] show_stack+0x40/0x128
[ 0.364289] [<8002fd88>] __warn+0xb8/0xe0
[ 0.364478] [<8002fe14>] warn_slowpath_fmt+0x64/0xc0
[ 0.364678] [<8072b1c8>] sched_clock_register+0x7c/0x2e4
[ 0.364895] [<8073c874>] ingenic_ost_probe+0x224/0x248
[ 0.365090] [<803d5394>] platform_drv_probe+0x40/0x94
[ 0.365526] [<803d362c>] really_probe+0x104/0x374
[ 0.365743] [<803d3ff0>] device_driver_attach+0x78/0x80
[ 0.365938] [<803d4070>] __driver_attach+0x78/0x118
[ 0.366129] [<803d1700>] bus_for_each_dev+0x7c/0xc8
[ 0.366318] [<803d226c>] bus_add_driver+0x1bc/0x204
[ 0.366513] [<803d4878>] driver_register+0x84/0x14c
[ 0.366717] [<8073a144>] __platform_driver_probe+0x98/0x140
[ 0.366931] [<80724e38>] do_one_initcall+0x84/0x1b4
[ 0.367126] [<807250cc>] kernel_init_freeable+0x164/0x240
[ 0.367318] [<805df75c>] kernel_init+0x10/0xf8
[ 0.367510] [<8001542c>] ret_from_kernel_thread+0x14/0x1c
[ 0.367722] ---[ end trace 7fedf00408fa3bed ]---
[ 0.367985] sched_clock: 32 bits at 12MHz, resolution 83ns, wraps
every 178956970966ns

At kernel/time/sched_clock.c:179 there is:
WARN_ON(!irqs_disabled());

That is strange, no drivers is doing that and no warning is appearing.

Isn't missing a local_irq_disable in the code path in the stack above?

I think it comes to the fact that the other drivers are probed much earlier in the boot process, while this one is probed as a regular platform device driver.


+ local_irq_save(flags);
+ if (soc_info->is64bit)
+ sched_clock_register(ingenic_ost_read_cntl, 32, rate);
+ else
+ sched_clock_register(ingenic_ost_read_cnth, 32, rate);
+ local_irq_restore(flags);
+
+ return 0;
+}
+


--
<http://www.linaro.org/> Linaro.org â Open source software for ARM SoCs

Follow Linaro: <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog