* [PATCH 0/2] net: wireless: rtw88: Fix oops on probe errors
@ 2020-09-20 13:26 Andreas Färber
2020-09-20 13:26 ` [PATCH 1/2] rtw88: Fix probe error handling race with firmware loading Andreas Färber
2020-09-20 13:26 ` [PATCH 2/2] rtw88: Fix potential probe error handling race with wow " Andreas Färber
0 siblings, 2 replies; 4+ messages in thread
From: Andreas Färber @ 2020-09-20 13:26 UTC (permalink / raw)
To: Yan-Hsuan Chuang, Kalle Valo, linux-wireless
Cc: Chin-Yen Lee, David S . Miller, Jakub Kicinski, netdev,
linux-realtek-soc, linux-arm-kernel, linux-kernel,
Andreas Färber
Hello,
This mini-series fixes oopses in rtw88 device probe error handling,
resulting from asynchronous firmware loading.
Since there does not appear to be a public kernel API for canceling
scheduled or ongoing firmware loads, it seems we need to wait with
teardown until rtw88's callback was invoked and signals completion.
Found on RTD1296 arm64 SoC with experimental PCI host bridge driver
(https://github.com/afaerber/linux/commits/rtd1295-next) with a 4K
physical bar window, resulting in rtw_pci_setup_resource() failing,
or with non-implemented 4K remapping resulting in rtw_pci_read32()
returning 0xffffffff values and causing rtw_mac_power_on() to fail.
Cheers,
Andreas
Andreas Färber (2):
rtw88: Fix probe error handling race with firmware loading
rtw88: Fix potential probe error handling race with wow firmware
loading
drivers/net/wireless/realtek/rtw88/main.c | 5 +++++
1 file changed, 5 insertions(+)
--
2.28.0
^ permalink raw reply [flat|nested] 4+ messages in thread
* [PATCH 1/2] rtw88: Fix probe error handling race with firmware loading
2020-09-20 13:26 [PATCH 0/2] net: wireless: rtw88: Fix oops on probe errors Andreas Färber
@ 2020-09-20 13:26 ` Andreas Färber
2020-09-24 15:50 ` Kalle Valo
2020-09-20 13:26 ` [PATCH 2/2] rtw88: Fix potential probe error handling race with wow " Andreas Färber
1 sibling, 1 reply; 4+ messages in thread
From: Andreas Färber @ 2020-09-20 13:26 UTC (permalink / raw)
To: Yan-Hsuan Chuang, Kalle Valo, linux-wireless
Cc: Chin-Yen Lee, David S . Miller, Jakub Kicinski, netdev,
linux-realtek-soc, linux-arm-kernel, linux-kernel,
Andreas Färber, Stanislaw Gruszka, Brian Norris
In case of rtw8822be, a probe failure after successful rtw_core_init()
has been observed to occasionally lead to an oops from rtw_load_firmware_cb():
[ 3.924268] pci 0001:01:00.0: [10ec:b822] type 00 class 0xff0000
[ 3.930531] pci 0001:01:00.0: reg 0x10: [io 0x0000-0x00ff]
[ 3.936360] pci 0001:01:00.0: reg 0x18: [mem 0x00000000-0x0000ffff 64bit]
[ 3.944042] pci 0001:01:00.0: supports D1 D2
[ 3.948438] pci 0001:01:00.0: PME# supported from D0 D1 D2 D3hot D3cold
[ 3.957312] pci 0001:01:00.0: BAR 2: no space for [mem size 0x00010000 64bit]
[ 3.964645] pci 0001:01:00.0: BAR 2: failed to assign [mem size 0x00010000 64bit]
[ 3.972332] pci 0001:01:00.0: BAR 0: assigned [io 0x10000-0x100ff]
[ 3.986240] rtw_8822be 0001:01:00.0: enabling device (0000 -> 0001)
[ 3.992735] rtw_8822be 0001:01:00.0: failed to map pci memory
[ 3.998638] rtw_8822be 0001:01:00.0: failed to request pci io region
[ 4.005166] rtw_8822be 0001:01:00.0: failed to setup pci resources
[ 4.011580] rtw_8822be: probe of 0001:01:00.0 failed with error -12
[ 4.018827] cfg80211: Loading compiled-in X.509 certificates for regulatory database
[ 4.029121] cfg80211: Loaded X.509 cert 'sforshee: 00b28ddf47aef9cea7'
[ 4.050828] Unable to handle kernel paging request at virtual address edafeaac9607952c
[ 4.058975] Mem abort info:
[ 4.058980] ESR = 0x96000004
[ 4.058990] EC = 0x25: DABT (current EL), IL = 32 bits
[ 4.070353] SET = 0, FnV = 0
[ 4.073487] EA = 0, S1PTW = 0
[ 4.073501] dw-apb-uart 98007800.serial: forbid DMA for kernel console
[ 4.076723] Data abort info:
[ 4.086415] ISV = 0, ISS = 0x00000004
[ 4.087731] Freeing unused kernel memory: 1792K
[ 4.090391] CM = 0, WnR = 0
[ 4.098091] [edafeaac9607952c] address between user and kernel address ranges
[ 4.105418] Internal error: Oops: 96000004 [#1] PREEMPT SMP
[ 4.111129] Modules linked in:
[ 4.114275] CPU: 1 PID: 31 Comm: kworker/1:1 Not tainted 5.9.0-rc5-next-20200915+ #700
[ 4.122386] Hardware name: Realtek Saola EVB (DT)
[ 4.127223] Workqueue: events request_firmware_work_func
[ 4.132676] pstate: 60000005 (nZCv daif -PAN -UAO BTYPE=--)
[ 4.138393] pc : rtw_load_firmware_cb+0x54/0xbc
[ 4.143040] lr : request_firmware_work_func+0x44/0xb4
[ 4.148217] sp : ffff800010133d70
[ 4.151616] x29: ffff800010133d70 x28: 0000000000000000
[ 4.157069] x27: 0000000000000000 x26: 0000000000000000
[ 4.162520] x25: 0000000000000000 x24: 0000000000000000
[ 4.167971] x23: ffff00007ac21908 x22: ffff00007ebb2100
[ 4.173424] x21: ffff00007ad35880 x20: edafeaac96079504
[ 4.178877] x19: ffff00007ad35870 x18: 0000000000000000
[ 4.184328] x17: 00000000000044d8 x16: 0000000000004310
[ 4.189780] x15: 0000000000000800 x14: 00000000ef006305
[ 4.195231] x13: ffffffff00000000 x12: ffffffffffffffff
[ 4.200682] x11: 0000000000000020 x10: 0000000000000003
[ 4.206135] x9 : 0000000000000000 x8 : ffff00007e73f680
[ 4.211585] x7 : 0000000000000000 x6 : ffff80001119b588
[ 4.217036] x5 : ffff00007e649c80 x4 : ffff00007e649c80
[ 4.222487] x3 : ffff80001119b588 x2 : ffff8000108d1718
[ 4.227940] x1 : ffff800011bd5000 x0 : ffff00007ac21600
[ 4.233391] Call trace:
[ 4.235906] rtw_load_firmware_cb+0x54/0xbc
[ 4.240198] request_firmware_work_func+0x44/0xb4
[ 4.245027] process_one_work+0x178/0x1e4
[ 4.249142] worker_thread+0x1d0/0x268
[ 4.252989] kthread+0xe8/0xf8
[ 4.256127] ret_from_fork+0x10/0x18
[ 4.259800] Code: f94013f5 a8c37bfd d65f03c0 f9000260 (f9401681)
[ 4.266049] ---[ end trace f822ebae1a8545c2 ]---
To avoid this, wait on the completion callbacks in rtw_core_deinit()
before releasing firmware and continuing teardown.
Note that rtw_wait_firmware_completion() was introduced with
c8e5695eae9959fc5774c0f490f2450be8bad3de ("rtw88: load wowlan firmware
if wowlan is supported"), so backports to earlier branches may need to
inline wait_for_completion(&rtwdev->fw.completion) instead.
Fixes: e3037485c68e ("rtw88: new Realtek 802.11ac driver")
Fixes: c8e5695eae99 ("rtw88: load wowlan firmware if wowlan is supported")
Cc: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
---
drivers/net/wireless/realtek/rtw88/main.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/net/wireless/realtek/rtw88/main.c b/drivers/net/wireless/realtek/rtw88/main.c
index 9770982b2f14..dc48ec4b0a31 100644
--- a/drivers/net/wireless/realtek/rtw88/main.c
+++ b/drivers/net/wireless/realtek/rtw88/main.c
@@ -1486,6 +1486,8 @@ void rtw_core_deinit(struct rtw_dev *rtwdev)
struct rtw_rsvd_page *rsvd_pkt, *tmp;
unsigned long flags;
+ rtw_wait_firmware_completion(rtwdev);
+
if (fw->firmware)
release_firmware(fw->firmware);
--
2.28.0
^ permalink raw reply related [flat|nested] 4+ messages in thread
* [PATCH 2/2] rtw88: Fix potential probe error handling race with wow firmware loading
2020-09-20 13:26 [PATCH 0/2] net: wireless: rtw88: Fix oops on probe errors Andreas Färber
2020-09-20 13:26 ` [PATCH 1/2] rtw88: Fix probe error handling race with firmware loading Andreas Färber
@ 2020-09-20 13:26 ` Andreas Färber
1 sibling, 0 replies; 4+ messages in thread
From: Andreas Färber @ 2020-09-20 13:26 UTC (permalink / raw)
To: Yan-Hsuan Chuang, Kalle Valo, linux-wireless
Cc: Chin-Yen Lee, David S . Miller, Jakub Kicinski, netdev,
linux-realtek-soc, linux-arm-kernel, linux-kernel,
Andreas Färber
If rtw_core_init() fails to load the wow firmware, rtw_core_deinit()
will not get called to clean up the regular firmware.
Ensure that an error loading the wow firmware does not produce an oops
for the regular firmware by waiting on its completion to be signalled
before returning. Also release the loaded firmware.
Fixes: c8e5695eae99 ("rtw88: load wowlan firmware if wowlan is supported")
Cc: Chin-Yen Lee <timlee@realtek.com>
Cc: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
---
drivers/net/wireless/realtek/rtw88/main.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/drivers/net/wireless/realtek/rtw88/main.c b/drivers/net/wireless/realtek/rtw88/main.c
index dc48ec4b0a31..cc82c80f0433 100644
--- a/drivers/net/wireless/realtek/rtw88/main.c
+++ b/drivers/net/wireless/realtek/rtw88/main.c
@@ -1472,6 +1472,9 @@ int rtw_core_init(struct rtw_dev *rtwdev)
ret = rtw_load_firmware(rtwdev, RTW_WOWLAN_FW);
if (ret) {
rtw_warn(rtwdev, "no wow firmware loaded\n");
+ wait_for_completion(&rtwdev->fw.completion);
+ if (rtwdev->fw.firmware)
+ release_firmware(rtwdev->fw.firmware);
return ret;
}
}
--
2.28.0
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH 1/2] rtw88: Fix probe error handling race with firmware loading
2020-09-20 13:26 ` [PATCH 1/2] rtw88: Fix probe error handling race with firmware loading Andreas Färber
@ 2020-09-24 15:50 ` Kalle Valo
0 siblings, 0 replies; 4+ messages in thread
From: Kalle Valo @ 2020-09-24 15:50 UTC (permalink / raw)
To: Andreas Färber
Cc: Yan-Hsuan Chuang, linux-wireless, Chin-Yen Lee, David S . Miller,
Jakub Kicinski, netdev, linux-realtek-soc, linux-arm-kernel,
linux-kernel, Andreas Färber, Stanislaw Gruszka,
Brian Norris
Andreas Färber <afaerber@suse.de> wrote:
> In case of rtw8822be, a probe failure after successful rtw_core_init()
> has been observed to occasionally lead to an oops from rtw_load_firmware_cb():
>
> [ 3.924268] pci 0001:01:00.0: [10ec:b822] type 00 class 0xff0000
> [ 3.930531] pci 0001:01:00.0: reg 0x10: [io 0x0000-0x00ff]
> [ 3.936360] pci 0001:01:00.0: reg 0x18: [mem 0x00000000-0x0000ffff 64bit]
> [ 3.944042] pci 0001:01:00.0: supports D1 D2
> [ 3.948438] pci 0001:01:00.0: PME# supported from D0 D1 D2 D3hot D3cold
> [ 3.957312] pci 0001:01:00.0: BAR 2: no space for [mem size 0x00010000 64bit]
> [ 3.964645] pci 0001:01:00.0: BAR 2: failed to assign [mem size 0x00010000 64bit]
> [ 3.972332] pci 0001:01:00.0: BAR 0: assigned [io 0x10000-0x100ff]
> [ 3.986240] rtw_8822be 0001:01:00.0: enabling device (0000 -> 0001)
> [ 3.992735] rtw_8822be 0001:01:00.0: failed to map pci memory
> [ 3.998638] rtw_8822be 0001:01:00.0: failed to request pci io region
> [ 4.005166] rtw_8822be 0001:01:00.0: failed to setup pci resources
> [ 4.011580] rtw_8822be: probe of 0001:01:00.0 failed with error -12
> [ 4.018827] cfg80211: Loading compiled-in X.509 certificates for regulatory database
> [ 4.029121] cfg80211: Loaded X.509 cert 'sforshee: 00b28ddf47aef9cea7'
> [ 4.050828] Unable to handle kernel paging request at virtual address edafeaac9607952c
> [ 4.058975] Mem abort info:
> [ 4.058980] ESR = 0x96000004
> [ 4.058990] EC = 0x25: DABT (current EL), IL = 32 bits
> [ 4.070353] SET = 0, FnV = 0
> [ 4.073487] EA = 0, S1PTW = 0
> [ 4.073501] dw-apb-uart 98007800.serial: forbid DMA for kernel console
> [ 4.076723] Data abort info:
> [ 4.086415] ISV = 0, ISS = 0x00000004
> [ 4.087731] Freeing unused kernel memory: 1792K
> [ 4.090391] CM = 0, WnR = 0
> [ 4.098091] [edafeaac9607952c] address between user and kernel address ranges
> [ 4.105418] Internal error: Oops: 96000004 [#1] PREEMPT SMP
> [ 4.111129] Modules linked in:
> [ 4.114275] CPU: 1 PID: 31 Comm: kworker/1:1 Not tainted 5.9.0-rc5-next-20200915+ #700
> [ 4.122386] Hardware name: Realtek Saola EVB (DT)
> [ 4.127223] Workqueue: events request_firmware_work_func
> [ 4.132676] pstate: 60000005 (nZCv daif -PAN -UAO BTYPE=--)
> [ 4.138393] pc : rtw_load_firmware_cb+0x54/0xbc
> [ 4.143040] lr : request_firmware_work_func+0x44/0xb4
> [ 4.148217] sp : ffff800010133d70
> [ 4.151616] x29: ffff800010133d70 x28: 0000000000000000
> [ 4.157069] x27: 0000000000000000 x26: 0000000000000000
> [ 4.162520] x25: 0000000000000000 x24: 0000000000000000
> [ 4.167971] x23: ffff00007ac21908 x22: ffff00007ebb2100
> [ 4.173424] x21: ffff00007ad35880 x20: edafeaac96079504
> [ 4.178877] x19: ffff00007ad35870 x18: 0000000000000000
> [ 4.184328] x17: 00000000000044d8 x16: 0000000000004310
> [ 4.189780] x15: 0000000000000800 x14: 00000000ef006305
> [ 4.195231] x13: ffffffff00000000 x12: ffffffffffffffff
> [ 4.200682] x11: 0000000000000020 x10: 0000000000000003
> [ 4.206135] x9 : 0000000000000000 x8 : ffff00007e73f680
> [ 4.211585] x7 : 0000000000000000 x6 : ffff80001119b588
> [ 4.217036] x5 : ffff00007e649c80 x4 : ffff00007e649c80
> [ 4.222487] x3 : ffff80001119b588 x2 : ffff8000108d1718
> [ 4.227940] x1 : ffff800011bd5000 x0 : ffff00007ac21600
> [ 4.233391] Call trace:
> [ 4.235906] rtw_load_firmware_cb+0x54/0xbc
> [ 4.240198] request_firmware_work_func+0x44/0xb4
> [ 4.245027] process_one_work+0x178/0x1e4
> [ 4.249142] worker_thread+0x1d0/0x268
> [ 4.252989] kthread+0xe8/0xf8
> [ 4.256127] ret_from_fork+0x10/0x18
> [ 4.259800] Code: f94013f5 a8c37bfd d65f03c0 f9000260 (f9401681)
> [ 4.266049] ---[ end trace f822ebae1a8545c2 ]---
>
> To avoid this, wait on the completion callbacks in rtw_core_deinit()
> before releasing firmware and continuing teardown.
>
> Note that rtw_wait_firmware_completion() was introduced with
> c8e5695eae9959fc5774c0f490f2450be8bad3de ("rtw88: load wowlan firmware
> if wowlan is supported"), so backports to earlier branches may need to
> inline wait_for_completion(&rtwdev->fw.completion) instead.
>
> Fixes: e3037485c68e ("rtw88: new Realtek 802.11ac driver")
> Fixes: c8e5695eae99 ("rtw88: load wowlan firmware if wowlan is supported")
> Cc: Yan-Hsuan Chuang <yhchuang@realtek.com>
> Signed-off-by: Andreas Färber <afaerber@suse.de>
2 patches applied to wireless-drivers-next.git, thanks.
ecda9cda3338 rtw88: Fix probe error handling race with firmware loading
ac4bac99161e rtw88: Fix potential probe error handling race with wow firmware loading
--
https://patchwork.kernel.org/patch/11787655/
https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2020-09-24 15:51 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-09-20 13:26 [PATCH 0/2] net: wireless: rtw88: Fix oops on probe errors Andreas Färber
2020-09-20 13:26 ` [PATCH 1/2] rtw88: Fix probe error handling race with firmware loading Andreas Färber
2020-09-24 15:50 ` Kalle Valo
2020-09-20 13:26 ` [PATCH 2/2] rtw88: Fix potential probe error handling race with wow " Andreas Färber
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).