RE: [PATCH v5 14/14] ASoC: rsnd: Add system suspend/resume support

[John Madieu]

Hi Geert,

Thanks for your review.

> -----Original Message-----
> From: Geert Uytterhoeven <geert@xxxxxxxxxxxxxx>
> Sent: Dienstag, 28. April 2026 12:25
> To: John Madieu <john.madieu@xxxxxxxxx>
> Subject: Re: [PATCH v5 14/14] ASoC: rsnd: Add system suspend/resume
> support
> 
> Hi John,
> 
> On Wed, 15 Apr 2026 at 14:47, John Madieu <john.madieu@xxxxxxxxx> wrote:
> > Add per-module suspend/resume functions following the existing driver
> > architecture where each module manages its own resources in its own
> > file. core.c provides common clock/reset helpers and orchestrates the
> > calls in the correct order (reverse probe for suspend, probe order for
> > resume).
> >
> > Infrastructure clocks (ADG, audmacpp, SCU) are managed globally using
> > optional APIs to remain transparent to platforms that don't specify
> > these clocks/resets.
> >
> > Signed-off-by: John Madieu <john.madieu.xa@xxxxxxxxxxxxxx>
> 
> Thanks for your patch!
> 
> This causes a crash on Gray Hawk Single (R-Car V4M), but not on Salvator-
> XS (R-Car H3 ES2.0):
> 
>     Unable to handle kernel NULL pointer dereference at virtual address
> 0000000000000008
>     Mem abort info:
>       ESR = 0x0000000096000004
>       EC = 0x25: DABT (current EL), IL = 32 bits
>       SET = 0, FnV = 0
>       EA = 0, S1PTW = 0
>       FSC = 0x04: level 0 translation fault
>     Data abort info:
>       ISV = 0, ISS = 0x00000004, ISS2 = 0x00000000
>       CM = 0, WnR = 0, TnD = 0, TagAccess = 0
>       GCS = 0, Overlay = 0, DirtyBit = 0, Xs = 0
>     user pgtable: 4k pages, 48-bit VAs, pgdp=00000004879ed000
>     [0000000000000008] pgd=0000000000000000, p4d=0000000000000000
>     Internal error: Oops: 0000000096000004 [#1]  SMP
>     CPU: 0 UID: 0 PID: 675 Comm: s2idle Not tainted
> 7.1.0-rc1-arm64-renesas-03756-g4cf47bc1e3d7 #3579 PREEMPT
>     Hardware name: Renesas Gray Hawk Single board based on r8a779h0 (DT)
>     pstate: 60400005 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
>     pc : rsnd_src_suspend+0x60/0x98
>     lr : rsnd_suspend+0x44/0x6c
>     sp : ffff800082e03ab0
>     x29: ffff800082e03ab0 x28: ffff000444022ac0 x27: ffff8000817d9050
>     x26: 0000000000000002 x25: ffff80008073dd64 x24: ffff800080ab0b20
>     x23: 0000000000000000 x22: ffff0004408a84a0 x21: 0000000000000002
>     x20: ffff0004402d6e80 x19: ffff0004402d6e80 x18: 0000000000000000
>     x17: 003fffffffffffff x16: ffff800082d8bc18 x15: ffff0005bef86f00
>     x14: ffff000444022b40 x13: 0000000000000112 x12: 071c71c71c71c71c
>     x11: 00000000000000c0 x10: 00000000000009f0 x9 : ffff800082e03880
>     x8 : ffff000444022ac0 x7 : ffff0005bef86f80 x6 : 0000000000000000
>     x5 : 0000000000000084 x4 : ffff000444022ac0 x3 : 0000000000000000
>     x2 : ffff000444022ac0 x1 : 0000000000000000 x0 : 0000000000000000
>     Call trace:
>      rsnd_src_suspend+0x60/0x98 (P)
>      rsnd_suspend+0x44/0x6c
>      device_suspend+0x134/0x3d8
>      dpm_suspend+0x164/0x268
>      dpm_suspend_start+0x4c/0x64
>      suspend_devices_and_enter+0x1c8/0x544
>      pm_suspend+0x194/0x1e0
>      state_store+0x7c/0x100
>      kobj_attr_store+0x14/0x24
>      sysfs_kf_write+0x78/0x8c
>      kernfs_fop_write_iter+0x128/0x1d0
>      vfs_write+0x210/0x390
>      ksys_write+0x6c/0x100
>      __arm64_sys_write+0x18/0x20
>      invoke_syscall+0x50/0x104
>      el0_svc_common.constprop.0+0x3c/0xd4
>      do_el0_svc+0x18/0x20
>      el0_svc+0x24/0xd8
>      el0t_64_sync_handler+0x98/0xdc
>      el0t_64_sync+0x154/0x158
>     Code: 6b13001f 54ffff0c f94013f5 f9402e80 (f9400413)
>     ---[ end trace 0000000000000000 ]---
> 
> > --- a/sound/soc/renesas/rcar/src.c
> > +++ b/sound/soc/renesas/rcar/src.c
> > @@ -848,3 +848,29 @@ void rsnd_src_remove(struct rsnd_priv *priv)
> >                 rsnd_mod_quit(rsnd_mod_get(src));
> >         }
> >  }
> > +
> > +void rsnd_src_suspend(struct rsnd_priv *priv) {
> > +       struct rsnd_src *src;
> > +       int i;
> > +
> > +       for_each_rsnd_src(src, priv, i)
> > +               rsnd_suspend_clk_reset(rsnd_mod_get(src)->clk,
> > +                                      rsnd_mod_get(src)->rstc);
> > +
> > +       clk_disable_unprepare(rsnd_priv_to_src_ctrl(priv)->scu_x2);
> 
> Apparently rsnd_priv_to_src_ctrl(priv) is NULL.

Indeed. Thanks for pointing it. In rsnd_src_probe(), we have:

	node = rsnd_src_of_node(priv);
	if (!node)
		return 0;

early in the function, so priv->src_ctrl stays NULL on platforms without SRC.
This should be fixed in v6 with proper checks in suspend/resume.

Regards,
John