drivers/pci/controller/dwc/pcie-rcar-gen4.c | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+)
The R-Car Gen4 PCIe controller is part of a power domain. On R-Car S4
and V4H, this is an always-on power domain which is not shut down in
suspend. On R-Car V4M, this is a dedicated A2PCIPHY power domain,
which is shut down during suspend, and the controller loses state,
which prevents the PCIe from working after resume.
Fix this by adding generic suspend/resume noirq ops for the controller,
which tear the link down on suspend, and restart it on resume. Use the
same PM ops on all of R-Car Gen4 to gracefully suspend and resume the
PCIe link on V4H and S4 too.
Test case which demonstrates the problem on R-Car V4M:
"
$ hexdump -vC /sys/bus/pci/devices/0000:00:00.0/config > /tmp/pre
$ echo s2idle > /sys/power/mem_sleep
$ echo platform > /sys/power/pm_test
$ echo mem > /sys/power/state
$ hexdump -vC /sys/bus/pci/devices/0000:00:00.0/config > /tmp/post
$ diff -Naru /tmp/pre /tmp/post
...
-00000000 12 19 32 00 07 05 10 00 00 00 04 06 00 00 01 00
+00000000 12 19 32 00 07 05 10 00 00 00 00 ff 00 00 80 00
^^^^^
Class 0604->00ff
"
Fixes: 0d0c551011df ("PCI: rcar-gen4: Add R-Car Gen4 PCIe controller support for host mode")
Reported-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
---
Cc: "Krzysztof Wilczyński" <kwilczynski@kernel.org>
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: Conor Dooley <conor+dt@kernel.org>
Cc: Geert Uytterhoeven <geert+renesas@glider.be>
Cc: Koichiro Den <den@valinux.co.jp>
Cc: Krzysztof Kozlowski <krzk+dt@kernel.org>
Cc: Lorenzo Pieralisi <lpieralisi@kernel.org>
Cc: Magnus Damm <magnus.damm@gmail.com>
Cc: Manivannan Sadhasivam <mani@kernel.org>
Cc: Rob Herring <robh@kernel.org>
Cc: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Cc: devicetree@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Cc: linux-pci@vger.kernel.org
Cc: linux-renesas-soc@vger.kernel.org
---
NOTE: Depends on series
https://lore.kernel.org/linux-pci/20260903205153.283553-1-marek.vasut+renesas@mailbox.org/
NOTE: I am sending this as an RFC, because I would like to get
a TB from Geert before this might get applied
---
drivers/pci/controller/dwc/pcie-rcar-gen4.c | 20 ++++++++++++++++++++
1 file changed, 20 insertions(+)
diff --git a/drivers/pci/controller/dwc/pcie-rcar-gen4.c b/drivers/pci/controller/dwc/pcie-rcar-gen4.c
index 5690dc32f47e0..3bfb2bd53fe3b 100644
--- a/drivers/pci/controller/dwc/pcie-rcar-gen4.c
+++ b/drivers/pci/controller/dwc/pcie-rcar-gen4.c
@@ -909,6 +909,22 @@ static int rcar_gen4_pcie_ltssm_control(struct rcar_gen4_pcie *rcar, bool enable
return 0;
}
+static int rcar_gen4_pcie_suspend_noirq(struct device *dev)
+{
+ struct rcar_gen4_pcie *rcar = dev_get_drvdata(dev);
+ struct dw_pcie *dw = &rcar->dw;
+
+ return dw_pcie_suspend_noirq(dw);
+}
+
+static int rcar_gen4_pcie_resume_noirq(struct device *dev)
+{
+ struct rcar_gen4_pcie *rcar = dev_get_drvdata(dev);
+ struct dw_pcie *dw = &rcar->dw;
+
+ return dw_pcie_resume_noirq(dw);
+}
+
static struct rcar_gen4_pcie_drvdata drvdata_r8a779f0_pcie = {
.ltssm_control = r8a779f0_pcie_ltssm_control,
.mode = DW_PCIE_RC_TYPE,
@@ -952,10 +968,14 @@ static const struct of_device_id rcar_gen4_pcie_of_match[] = {
};
MODULE_DEVICE_TABLE(of, rcar_gen4_pcie_of_match);
+DEFINE_NOIRQ_DEV_PM_OPS(rcar_gen4_pcie_pm_ops,
+ rcar_gen4_pcie_suspend_noirq, rcar_gen4_pcie_resume_noirq);
+
static struct platform_driver rcar_gen4_pcie_driver = {
.driver = {
.name = "pcie-rcar-gen4",
.of_match_table = rcar_gen4_pcie_of_match,
+ .pm = &rcar_gen4_pcie_pm_ops,
.probe_type = PROBE_PREFER_ASYNCHRONOUS,
},
.probe = rcar_gen4_pcie_probe,
--
2.53.0
Hi Marek,
On Sun, 6 Sept 2026 at 19:32, Marek Vasut
<marek.vasut+renesas@mailbox.org> wrote:
> The R-Car Gen4 PCIe controller is part of a power domain. On R-Car S4
> and V4H, this is an always-on power domain which is not shut down in
> suspend. On R-Car V4M, this is a dedicated A2PCIPHY power domain,
> which is shut down during suspend, and the controller loses state,
> which prevents the PCIe from working after resume.
>
> Fix this by adding generic suspend/resume noirq ops for the controller,
> which tear the link down on suspend, and restart it on resume. Use the
> same PM ops on all of R-Car Gen4 to gracefully suspend and resume the
> PCIe link on V4H and S4 too.
>
> Test case which demonstrates the problem on R-Car V4M:
> "
> $ hexdump -vC /sys/bus/pci/devices/0000:00:00.0/config > /tmp/pre
> $ echo s2idle > /sys/power/mem_sleep
> $ echo platform > /sys/power/pm_test
> $ echo mem > /sys/power/state
> $ hexdump -vC /sys/bus/pci/devices/0000:00:00.0/config > /tmp/post
> $ diff -Naru /tmp/pre /tmp/post
> ...
> -00000000 12 19 32 00 07 05 10 00 00 00 04 06 00 00 01 00
> +00000000 12 19 32 00 07 05 10 00 00 00 00 ff 00 00 80 00
> ^^^^^
> Class 0604->00ff
> "
>
> Fixes: 0d0c551011df ("PCI: rcar-gen4: Add R-Car Gen4 PCIe controller support for host mode")
> Reported-by: Geert Uytterhoeven <geert+renesas@glider.be>
> Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Thanks, the output of "lspci -v" is now the same before and after
s2idle.
Tested-by: Geert Uytterhoeven <geert+renesas@glider.be>
I don't have a setup for endpoint mode, so I could not verify or refute
Sashiko's claim that suspend will crash in endpoint mode.
> --- a/drivers/pci/controller/dwc/pcie-rcar-gen4.c
> +++ b/drivers/pci/controller/dwc/pcie-rcar-gen4.c
> @@ -952,10 +968,14 @@ static const struct of_device_id rcar_gen4_pcie_of_match[] = {
> };
> MODULE_DEVICE_TABLE(of, rcar_gen4_pcie_of_match);
>
> +DEFINE_NOIRQ_DEV_PM_OPS(rcar_gen4_pcie_pm_ops,
> + rcar_gen4_pcie_suspend_noirq, rcar_gen4_pcie_resume_noirq);
> +
> static struct platform_driver rcar_gen4_pcie_driver = {
> .driver = {
> .name = "pcie-rcar-gen4",
> .of_match_table = rcar_gen4_pcie_of_match,
> + .pm = &rcar_gen4_pcie_pm_ops,
pm_sleep_ptr(&rcar_gen4_pcie_pm_ops)
> .probe_type = PROBE_PREFER_ASYNCHRONOUS,
> },
> .probe = rcar_gen4_pcie_probe,
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
On 9/7/26 12:26 PM, Geert Uytterhoeven wrote:
Hello Geert,
> On Sun, 6 Sept 2026 at 19:32, Marek Vasut
> <marek.vasut+renesas@mailbox.org> wrote:
>> The R-Car Gen4 PCIe controller is part of a power domain. On R-Car S4
>> and V4H, this is an always-on power domain which is not shut down in
>> suspend. On R-Car V4M, this is a dedicated A2PCIPHY power domain,
>> which is shut down during suspend, and the controller loses state,
>> which prevents the PCIe from working after resume.
>>
>> Fix this by adding generic suspend/resume noirq ops for the controller,
>> which tear the link down on suspend, and restart it on resume. Use the
>> same PM ops on all of R-Car Gen4 to gracefully suspend and resume the
>> PCIe link on V4H and S4 too.
>>
>> Test case which demonstrates the problem on R-Car V4M:
>> "
>> $ hexdump -vC /sys/bus/pci/devices/0000:00:00.0/config > /tmp/pre
>> $ echo s2idle > /sys/power/mem_sleep
>> $ echo platform > /sys/power/pm_test
>> $ echo mem > /sys/power/state
>> $ hexdump -vC /sys/bus/pci/devices/0000:00:00.0/config > /tmp/post
>> $ diff -Naru /tmp/pre /tmp/post
>> ...
>> -00000000 12 19 32 00 07 05 10 00 00 00 04 06 00 00 01 00
>> +00000000 12 19 32 00 07 05 10 00 00 00 00 ff 00 00 80 00
>> ^^^^^
>> Class 0604->00ff
>> "
>>
>> Fixes: 0d0c551011df ("PCI: rcar-gen4: Add R-Car Gen4 PCIe controller support for host mode")
>> Reported-by: Geert Uytterhoeven <geert+renesas@glider.be>
>> Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
>
> Thanks, the output of "lspci -v" is now the same before and after
> s2idle.
> Tested-by: Geert Uytterhoeven <geert+renesas@glider.be>
>
> I don't have a setup for endpoint mode, so I could not verify or refute
> Sashiko's claim that suspend will crash in endpoint mode.
Same here (thus far).
>> --- a/drivers/pci/controller/dwc/pcie-rcar-gen4.c
>> +++ b/drivers/pci/controller/dwc/pcie-rcar-gen4.c
>
>> @@ -952,10 +968,14 @@ static const struct of_device_id rcar_gen4_pcie_of_match[] = {
>> };
>> MODULE_DEVICE_TABLE(of, rcar_gen4_pcie_of_match);
>>
>> +DEFINE_NOIRQ_DEV_PM_OPS(rcar_gen4_pcie_pm_ops,
>> + rcar_gen4_pcie_suspend_noirq, rcar_gen4_pcie_resume_noirq);
>> +
>> static struct platform_driver rcar_gen4_pcie_driver = {
>> .driver = {
>> .name = "pcie-rcar-gen4",
>> .of_match_table = rcar_gen4_pcie_of_match,
>> + .pm = &rcar_gen4_pcie_pm_ops,
>
> pm_sleep_ptr(&rcar_gen4_pcie_pm_ops)
Will do in V2.
On Mon, Sep 07, 2026 at 12:26:42PM +0200, Geert Uytterhoeven wrote:
> Hi Marek,
>
> On Sun, 6 Sept 2026 at 19:32, Marek Vasut
> <marek.vasut+renesas@mailbox.org> wrote:
> > The R-Car Gen4 PCIe controller is part of a power domain. On R-Car S4
> > and V4H, this is an always-on power domain which is not shut down in
> > suspend. On R-Car V4M, this is a dedicated A2PCIPHY power domain,
> > which is shut down during suspend, and the controller loses state,
> > which prevents the PCIe from working after resume.
> >
> > Fix this by adding generic suspend/resume noirq ops for the controller,
> > which tear the link down on suspend, and restart it on resume. Use the
> > same PM ops on all of R-Car Gen4 to gracefully suspend and resume the
> > PCIe link on V4H and S4 too.
> >
> > Test case which demonstrates the problem on R-Car V4M:
> > "
> > $ hexdump -vC /sys/bus/pci/devices/0000:00:00.0/config > /tmp/pre
> > $ echo s2idle > /sys/power/mem_sleep
> > $ echo platform > /sys/power/pm_test
> > $ echo mem > /sys/power/state
> > $ hexdump -vC /sys/bus/pci/devices/0000:00:00.0/config > /tmp/post
> > $ diff -Naru /tmp/pre /tmp/post
> > ...
> > -00000000 12 19 32 00 07 05 10 00 00 00 04 06 00 00 01 00
> > +00000000 12 19 32 00 07 05 10 00 00 00 00 ff 00 00 80 00
> > ^^^^^
> > Class 0604->00ff
> > "
> >
> > Fixes: 0d0c551011df ("PCI: rcar-gen4: Add R-Car Gen4 PCIe controller support for host mode")
> > Reported-by: Geert Uytterhoeven <geert+renesas@glider.be>
> > Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
>
> Thanks, the output of "lspci -v" is now the same before and after
> s2idle.
> Tested-by: Geert Uytterhoeven <geert+renesas@glider.be>
>
> I don't have a setup for endpoint mode, so I could not verify or refute
> Sashiko's claim that suspend will crash in endpoint mode.
Hi Marek, Geert,
I have reproduced the NULL pointer dereference at pci->pp.ops->deinit on R-Car
S4 Spider in EP mode with CONFIG_PCIE_DW_HOST=y.
(I used next-20260904 with this patch and the following fix applied:
"buffer: fix NULL dereference of bh->b_folio in __bh_submit()")
[ 102.467803] PM: suspend entry (s2idle)
[ 102.495431] Filesystems sync: 0.027 seconds
[ 102.499213] Freezing user space processes
[ 102.502243] Freezing user space processes completed (elapsed 0.002 seconds)
[ 102.502783] OOM killer disabled.
[ 102.503013] Freezing remaining freezable tasks
[ 102.505171] Freezing remaining freezable tasks completed (elapsed 0.001 seconds)
[ 102.508276] GFP mask restricted
[ 102.554124] renesas_eth_sw e6880000.ethernet tsn0: Link is Down
[ 102.557676] PM: suspend of devices complete after 49.149 msecs
[ 102.558098] PM: start suspend of devices complete after 52.385 msecs
[ 102.562380] PM: late suspend of devices complete after 3.853 msecs
[ 102.564497] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000008
[ 102.565156] Mem abort info:
[ 102.565344] ESR = 0x0000000096000004
[ 102.565591] EC = 0x25: DABT (current EL), IL = 32 bits
[ 102.565938] SET = 0, FnV = 0
[ 102.566140] EA = 0, S1PTW = 0
[ 102.566347] FSC = 0x04: level 0 translation fault
[ 102.566665] Data abort info:
[ 102.566856] ISV = 0, ISS = 0x00000004, ISS2 = 0x00000000
[ 102.567212] CM = 0, WnR = 0, TnD = 0, TagAccess = 0
[ 102.567541] GCS = 0, Overlay = 0, DirtyBit = 0
[ 102.567842] user pgtable: 4k pages, 48-bit VAs, pgdp=0000000487961000
[ 102.568260] [0000000000000008] pgd=0000000000000000, p4d=0000000000000000
[ 102.568747] Internal error: Oops: 0000000096000004 [#1] SMP
[ 102.569124] Modules linked in: sch_fq_codel
[ 102.569417] CPU: 1 UID: 0 PID: 1837 Comm: bash Not tainted 7.3.0-rc1-next-20260904-ep-00002-gb46bcc4e34bb #3 PREEMPT
[ 102.570107] Hardware name: Renesas Spider CPU and Breakout boards based on r8a779f0 (DT)
[ 102.570632] pstate: 40400009 (nZcv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[ 102.571086] pc : dw_pcie_suspend_noirq+0x5c/0x258
[ 102.571414] lr : dw_pcie_suspend_noirq+0x58/0x258
[ 102.571725] sp : ffff80008504b970
[ 102.571944] x29: ffff80008504b9b0 x28: 0000000000000000 x27: ffff800082baccd8
[ 102.572416] x26: ffff800082cddcf8 x25: ffff800082bacce8 x24: 0000000000000000
[ 102.572887] x23: ffff8000809f3b88 x22: ffff000440377310 x21: ffff800082d0f000
[ 102.573358] x20: ffff000440c58410 x19: ffff000442418080 x18: 000000000000000a
[ 102.573829] x17: 000000040044ffff x16: 041000f2b5503510 x15: 0000000000000000
[ 102.574299] x14: ffff000444ffbda0 x13: 00000000000001e1 x12: 0000000000000000
[ 102.574769] x11: 00000000000000c0 x10: 0000000000000af0 x9 : ffff8000808520cc
[ 102.575240] x8 : ffff8004397b2000 x7 : 0000000000000eae x6 : 00000000050f95a1
[ 102.575710] x5 : 0000000000000eaf x4 : fffffffffffff000 x3 : 0000000000000001
[ 102.576180] x2 : 0000000000010000 x1 : 0000000000000000 x0 : 0000000000000000
[ 102.576651] Call trace:
[ 102.576818] dw_pcie_suspend_noirq+0x5c/0x258 (P)
[ 102.577133] rcar_gen4_pcie_suspend_noirq+0x18/0x28
[ 102.577457] pm_generic_suspend_noirq+0x30/0x50
[ 102.577764] genpd_finish_suspend+0x48/0x178
[ 102.578053] genpd_suspend_noirq+0x30/0x90
[ 102.578326] dpm_run_callback+0x5c/0x2f8
[ 102.578590] device_suspend_noirq+0x74/0x260
[ 102.578874] dpm_noirq_suspend_devices+0x190/0x2a8
[ 102.579192] dpm_suspend_noirq+0x28/0xa8
[ 102.579455] suspend_devices_and_enter+0x310/0x6e8
[ 102.579778] pm_suspend+0x328/0x3f8
[ 102.580012] state_store+0x84/0x110
[ 102.580247] kobj_attr_store+0x14/0x28
[ 102.580504] sysfs_kf_write+0x80/0xa0
[ 102.580758] kernfs_fop_write_iter+0x12c/0x200
[ 102.581055] vfs_write+0x1f8/0x388
[ 102.581289] ksys_write+0x64/0x108
[ 102.581517] __arm64_sys_write+0x1c/0x38
[ 102.581780] invoke_syscall.constprop.0+0x44/0x110
[ 102.582102] el0_svc_common.constprop.0+0x3c/0xe0
[ 102.582415] do_el0_svc+0x20/0x30
[ 102.582640] el0_svc+0x34/0x1c8
[ 102.582855] el0t_64_sync_handler+0x98/0xe0
[ 102.583134] el0t_64_sync+0x170/0x178
[ 102.583388] Code: b4000061 aa1303e0 d63f0020 f9404e60 (f9400401)
[ 102.583793] ---[ end trace 0000000000000000 ]---
[ 102.584102] Kernel panic - not syncing: Oops: Fatal exception
[ 102.584479] SMP: stopping secondary CPUs
[ 102.584822] Kernel Offset: disabled
[ 102.585052] CPU features: 0x0,40000000,0040002a,01180a42,0800e01b
[ 102.585451] Memory Limit: none
[ 102.585659] Rebooting in 30 seconds..
[drivers/pci/controller/dwc/pcie-designware-host.c]
1225 int dw_pcie_suspend_noirq(struct dw_pcie *pci)
1226 {
...
1286 pci->pp.skip_pwrctrl_off = pme_capable;
1287 dw_pcie_stop_link(pci);
> 1288 if (pci->pp.ops->deinit)
1289 pci->pp.ops->deinit(&pci->pp);
1290
Best regards,
Koichiro
>
> > --- a/drivers/pci/controller/dwc/pcie-rcar-gen4.c
> > +++ b/drivers/pci/controller/dwc/pcie-rcar-gen4.c
>
> > @@ -952,10 +968,14 @@ static const struct of_device_id rcar_gen4_pcie_of_match[] = {
> > };
> > MODULE_DEVICE_TABLE(of, rcar_gen4_pcie_of_match);
> >
> > +DEFINE_NOIRQ_DEV_PM_OPS(rcar_gen4_pcie_pm_ops,
> > + rcar_gen4_pcie_suspend_noirq, rcar_gen4_pcie_resume_noirq);
> > +
> > static struct platform_driver rcar_gen4_pcie_driver = {
> > .driver = {
> > .name = "pcie-rcar-gen4",
> > .of_match_table = rcar_gen4_pcie_of_match,
> > + .pm = &rcar_gen4_pcie_pm_ops,
>
> pm_sleep_ptr(&rcar_gen4_pcie_pm_ops)
>
> > .probe_type = PROBE_PREFER_ASYNCHRONOUS,
> > },
> > .probe = rcar_gen4_pcie_probe,
>
> Gr{oetje,eeting}s,
>
> Geert
>
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
>
> In personal conversations with technical people, I call myself a hacker. But
> when I'm talking to journalists I just say "programmer" or something like that.
> -- Linus Torvalds
On 9/7/26 3:46 PM, Koichiro Den wrote:
Hello Den-san,
> I have reproduced the NULL pointer dereference at pci->pp.ops->deinit on R-Car
> S4 Spider in EP mode with CONFIG_PCIE_DW_HOST=y.
>
> (I used next-20260904 with this patch and the following fix applied:
> "buffer: fix NULL dereference of bh->b_folio in __bh_submit()")
[...]
> [drivers/pci/controller/dwc/pcie-designware-host.c]
> 1225 int dw_pcie_suspend_noirq(struct dw_pcie *pci)
> 1226 {
> ...
> 1286 pci->pp.skip_pwrctrl_off = pme_capable;
> 1287 dw_pcie_stop_link(pci);
> > 1288 if (pci->pp.ops->deinit)
> 1289 pci->pp.ops->deinit(&pci->pp);
> 1290
Thank you for your help with the endpoint part.
The fix turns out to be very simple -- the suspend/resume handling is
specific to the PCIe controller in RC mode, so I added a conditional to
invoke those host (RC) suspend/resume operations only if the controller
is in that mode, and I sent a V2.
This RC-only fix in V2 is also in line with what MX6 and DRA7xx PCIe
controllers do.
--
Best regards,
Marek Vasut
© 2016 - 2026 Red Hat, Inc.