Re: [PATCH] clk: vc5: Fix div-by-0 when rounding a rate of zero

From: Steve Longerbeam
Date: Thu May 31 2018 - 14:32:10 EST


Hi Marek,


On 05/31/2018 11:25 AM, Marek Vasut wrote:
On 05/31/2018 08:23 PM, Steve Longerbeam wrote:
Just return zero for a rounded rate if requested rate is zero.

This was caught by CONFIG_UBSAN:

[ 192.266748] UBSAN: Undefined behaviour in drivers/clk/clk-versaclock5.c:513:17
[ 192.274050] division by zero
[ 192.276976] CPU: 0 PID: 2579 Comm: vsp-unit-test-0 Tainted: G B WC 4.14.17-02752-g13fb96f #1
[ 192.286378] Hardware name: Renesas Salvator-X board based on r8a7795 ES2.0+ (DT)
[ 192.293852] Call trace:
[ 192.296343] [<ffff2000080900dc>] dump_backtrace+0x0/0x390
[ 192.301807] [<ffff200008090480>] show_stack+0x14/0x1c
[ 192.306920] [<ffff200008f66574>] dump_stack+0x134/0x1a8
[ 192.312213] [<ffff2000087aaa30>] ubsan_epilogue+0x14/0x60
[ 192.317677] [<ffff2000087ab4d0>] __ubsan_handle_divrem_overflow+0x11c/0x170
[ 192.324720] [<ffff200008852120>] vc5_fod_round_rate+0x68/0x148
[ 192.330620] [<ffff20000884567c>] clk_calc_new_rates+0x238/0x3fc
[ 192.336607] [<ffff2000088456e0>] clk_calc_new_rates+0x29c/0x3fc
[ 192.342595] [<ffff2000088483ac>] clk_core_set_rate_nolock+0x48/0x11c
[ 192.349019] [<ffff2000088484b4>] clk_set_rate+0x34/0x4c
[ 192.354307] [<ffff20000895e304>] rcar_du_pm_suspend+0x274/0x2f4
[ 192.360297] [<ffff20000898feac>] platform_pm_suspend+0x78/0xb8
[ 192.366198] [<ffff2000089a5604>] dpm_run_callback+0x584/0xa18
[ 192.372010] [<ffff2000089a69e0>] __device_suspend+0x1a8/0x534
[ 192.377822] [<ffff2000089adc48>] dpm_suspend+0x130/0xea0
[ 192.383197] [<ffff2000089b0344>] dpm_suspend_start+0x130/0x138
[ 192.389099] [<ffff20000817f584>] suspend_devices_and_enter+0xf0/0x1778
[ 192.395700] [<ffff200008183014>] pm_suspend+0x2408/0x245c
[ 192.401162] [<ffff20000817c0a4>] state_store+0xf0/0x130
[ 192.406451] [<ffff200008f6f19c>] kobj_attr_store+0x5c/0x6c
[ 192.412002] [<ffff2000084f4c94>] sysfs_kf_write+0xe8/0xfc
[ 192.417466] [<ffff2000084f30b0>] kernfs_fop_write+0x22c/0x2e4
[ 192.423281] [<ffff2000083e46d4>] __vfs_write+0x104/0x34c
[ 192.428656] [<ffff2000083e4cc4>] vfs_write+0x134/0x2d8
[ 192.433857] [<ffff2000083e5150>] SyS_write+0xbc/0x12c
[ 192.438967] Exception stack(0xffff8006cd1cfec0 to 0xffff8006cd1d0000)
[ 192.445480] fec0: 0000000000000001 000000001e303f00 0000000000000004 0000ffff959a5000
[ 192.453397] fee0: 0000000000000000 0000000155510004 0000000000000003 000000000000006d
[ 192.461314] ff00: 0000000000000040 0000000000000000 0000ffffcc304800 0000000000000020
[ 192.469230] ff20: 0000000000000000 0000000000000000 0000000000000001 0000000000000008
[ 192.477148] ff40: 00000000004eb3b8 0000ffff958bb840 000000000000003d 0000000000000001
[ 192.485065] ff60: 000000001e303f00 0000ffff959a1508 0000000000000004 000000001e303f00
[ 192.492982] ff80: 0000000000000004 00000000004d4c68 0000000000000001 0000000000000000
[ 192.500899] ffa0: 000000001e30d5c0 0000ffffcc304820 0000ffff958bec64 0000ffffcc304820
[ 192.508816] ffc0: 0000ffff95912898 0000000020000000 0000000000000001 0000000000000040
[ 192.516733] ffe0: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
[ 192.524650] [<ffff200008083ef0>] el0_svc_naked+0x24/0x28

Signed-off-by: Steve Longerbeam <steve_longerbeam@xxxxxxxxxx>
---
drivers/clk/clk-versaclock5.c | 4 ++++
1 file changed, 4 insertions(+)

diff --git a/drivers/clk/clk-versaclock5.c b/drivers/clk/clk-versaclock5.c
index decffb3..113523d 100644
--- a/drivers/clk/clk-versaclock5.c
+++ b/drivers/clk/clk-versaclock5.c
@@ -509,6 +509,10 @@ static long vc5_fod_round_rate(struct clk_hw *hw, unsigned long rate,
u32 div_int;
u64 div_frc;
+ /* prevent div-by-0 */
+ if (rate == 0)
+ return 0;
+
/* Determine integer part, which is 12 bit wide */
div_int = f_in / rate;
/*

Can this actually happen ?

We caught this using the Renesas 3.6.0 BSP release, when performing
a suspend of rcar-du driver. The rcar_du_pm_suspend() in 3.6.0 BSP is modified
from mainline version, including calling clk_set_rate() on the crtc clocks with a
rate of zero. So this is not actually reproducible (yet) in mainline.

Steve