Hi,
With the rtl8192ce driver, I get freezes a few seconds after connecting
to a network. (see PCI id below) Machine is unusable after hang. I'm
using the linville/wireless-next git tree.
I've been playing with the different options, and noticed that setting
fwlps=0 fixes the issue.
I've tried to get more messages with netconsole, but once the lockup
happens I'm unable to get any debug message (regardless of the log level
of the driver).
Regards,
Anisse
06:00.0 Network controller [0280]: Realtek Semiconductor Co., Ltd. RTL8188CE 802.11b/g/n WiFi Adapter [10ec:8176] (rev 01)
Subsystem: Realtek Semiconductor Co., Ltd. Device [10ec:8212]
Flags: bus master, fast devsel, latency 0, IRQ 17
I/O ports at 3000 [size=256]
Memory at f0200000 (64-bit, non-prefetchable) [size=16K]
Capabilities: [40] Power Management version 3
Capabilities: [50] MSI: Enable- Count=1/1 Maskable- 64bit+
Capabilities: [70] Express Endpoint, MSI 00
Capabilities: [100] Advanced Error Reporting
Capabilities: [140] Virtual Channel
Capabilities: [160] Device Serial Number 01-91-81-fe-ff-4c-e0-00
Kernel driver in use: rtl8192ce
On 05/16/2012 05:41 AM, Anisse Astier wrote:
> I enabled the following options:
> CONFIG_LOCKDEP_SUPPORT=y
> CONFIG_LOCKDEP=y
> CONFIG_DEBUG_LOCKDEP=y
> CONFIG_DEBUG_SPINLOCK=y
> CONFIG_DEBUG_MUTEXES=y
> But it didn't detect any lockup yet, and machine still hangs. I'll try
> CONFIG_PROVE_LOCKING.
>
> Enabling CONFIG_DEBUG_INFO seems to make the problem go away...
That may explain why I never see the problem. I do not think I have ever
generated a kernel with that parameter disabled, and my distro's kernels have it
enabled. Does anything new get logged (dmesg output) when that parameter is enabled?
Larry
On 05/15/2012 10:04 AM, Anisse Astier wrote:
> On Tue, 15 May 2012 09:51:23 -0500, Larry Finger<[email protected]> wrote :
>
> I'll try wireless-testing then. Are all rtlwifi patches in
> wireless-testing ?
I think that all are in wireless-testing.
> (This problem was also reproduced with 3.2.X)
>
>
>> to say that I cannot duplicate your results, and that we need to explore the
>> rest of your system to see what is different.
>>
>> Is your system 32- or 64-bit? How many cores in your CPU? Have you tried
> 32 bits, SoC is an AMD E1-1200, with 2 cores.
My search shows that the clock rate for this processor is 1.4 GHz. The only
system that I have that can run the RTL8188CE is 2 GHz.
>> connecting to more than one AP? What is the make and model of the AP that needs
> I've tried two different APs, with and without keys (open and WPA2)
>
>> to have power save turned off? What is the mode of operation?
> One is a Linksys WRT54GL with dd-wrt(WPA2), the other(Open) is a
> carrier's set top box/router (Freebox).
I thought the locking in the Power Save routines was OK, but perhaps there is
some raciness that is exposed by the difference in speeds. The lack of
diagnostics on the netconsole is a problem.
Do you have kernel debugging turned on for the lock diagnostics, and the Detect
Hung Task configuration variable? Perhaps we can get some logging that way.
Have you tried the vendor's driver from http://www.realtek.com? I don't think it will
make a difference, but it could be worth a try.
Larry
On Tue, 15 May 2012 09:51:23 -0500, Larry Finger <[email protected]> wrote :
> On 05/15/2012 08:29 AM, Anisse Astier wrote:
> > Hi,
> >
> > With the rtl8192ce driver, I get freezes a few seconds after connecting
> > to a network. (see PCI id below) Machine is unusable after hang. I'm
> > using the linville/wireless-next git tree.
> >
> > I've been playing with the different options, and noticed that setting
> > fwlps=0 fixes the issue.
> >
> > I've tried to get more messages with netconsole, but once the lockup
> > happens I'm unable to get any debug message (regardless of the log level
> > of the driver).
> >
> > Regards,
> >
> > Anisse
> >
> >
> > 06:00.0 Network controller [0280]: Realtek Semiconductor Co., Ltd. RTL8188CE 802.11b/g/n WiFi Adapter [10ec:8176] (rev 01)
> > Subsystem: Realtek Semiconductor Co., Ltd. Device [10ec:8212]
> > Flags: bus master, fast devsel, latency 0, IRQ 17
> > I/O ports at 3000 [size=256]
> > Memory at f0200000 (64-bit, non-prefetchable) [size=16K]
> > Capabilities: [40] Power Management version 3
> > Capabilities: [50] MSI: Enable- Count=1/1 Maskable- 64bit+
> > Capabilities: [70] Express Endpoint, MSI 00
> > Capabilities: [100] Advanced Error Reporting
> > Capabilities: [140] Virtual Channel
> > Capabilities: [160] Device Serial Number 01-91-81-fe-ff-4c-e0-00
> > Kernel driver in use: rtl8192ce
>
> I have been doing long-term tests with that same card for about one week with no
> dropouts and no freezes. I am running wireless-testing, not wireless-next;
> however, I have all the rtlwifi and friends patches applied. I think it is safe
I'll try wireless-testing then. Are all rtlwifi patches in
wireless-testing ?
(This problem was also reproduced with 3.2.X)
> to say that I cannot duplicate your results, and that we need to explore the
> rest of your system to see what is different.
>
> Is your system 32- or 64-bit? How many cores in your CPU? Have you tried
32 bits, SoC is an AMD E1-1200, with 2 cores.
> connecting to more than one AP? What is the make and model of the AP that needs
I've tried two different APs, with and without keys (open and WPA2)
> to have power save turned off? What is the mode of operation?
One is a Linksys WRT54GL with dd-wrt(WPA2), the other(Open) is a
carrier's set top box/router (Freebox).
>
> I am using a 64-bit system with a dual-core CPU. My main AP is a Netgear
> WNDR3300 running standard firmware and is encrypted using WPA2-TKIP (AES). The
> mode is 802.11n at up to 270 Mbps; however, I have also tested using a Linksys
> WRT54GL running openWRT Kamakaze firmware. That one is 802.11g at up to 54 Mbps.
>
> Larry
Anisse
On 05/15/2012 08:29 AM, Anisse Astier wrote:
> Hi,
>
> With the rtl8192ce driver, I get freezes a few seconds after connecting
> to a network. (see PCI id below) Machine is unusable after hang. I'm
> using the linville/wireless-next git tree.
>
> I've been playing with the different options, and noticed that setting
> fwlps=0 fixes the issue.
>
> I've tried to get more messages with netconsole, but once the lockup
> happens I'm unable to get any debug message (regardless of the log level
> of the driver).
>
> Regards,
>
> Anisse
>
>
> 06:00.0 Network controller [0280]: Realtek Semiconductor Co., Ltd. RTL8188CE 802.11b/g/n WiFi Adapter [10ec:8176] (rev 01)
> Subsystem: Realtek Semiconductor Co., Ltd. Device [10ec:8212]
> Flags: bus master, fast devsel, latency 0, IRQ 17
> I/O ports at 3000 [size=256]
> Memory at f0200000 (64-bit, non-prefetchable) [size=16K]
> Capabilities: [40] Power Management version 3
> Capabilities: [50] MSI: Enable- Count=1/1 Maskable- 64bit+
> Capabilities: [70] Express Endpoint, MSI 00
> Capabilities: [100] Advanced Error Reporting
> Capabilities: [140] Virtual Channel
> Capabilities: [160] Device Serial Number 01-91-81-fe-ff-4c-e0-00
> Kernel driver in use: rtl8192ce
I have been doing long-term tests with that same card for about one week with no
dropouts and no freezes. I am running wireless-testing, not wireless-next;
however, I have all the rtlwifi and friends patches applied. I think it is safe
to say that I cannot duplicate your results, and that we need to explore the
rest of your system to see what is different.
Is your system 32- or 64-bit? How many cores in your CPU? Have you tried
connecting to more than one AP? What is the make and model of the AP that needs
to have power save turned off? What is the mode of operation?
I am using a 64-bit system with a dual-core CPU. My main AP is a Netgear
WNDR3300 running standard firmware and is encrypted using WPA2-TKIP (AES). The
mode is 802.11n at up to 270 Mbps; however, I have also tested using a Linksys
WRT54GL running openWRT Kamakaze firmware. That one is 802.11g at up to 54 Mbps.
Larry
On Tue, 15 May 2012 11:45:40 -0500, Larry Finger <[email protected]> wrote :
> On 05/15/2012 10:04 AM, Anisse Astier wrote:
> > On Tue, 15 May 2012 09:51:23 -0500, Larry Finger<[email protected]> wrote :
> >
> > I'll try wireless-testing then. Are all rtlwifi patches in
> > wireless-testing ?
>
> I think that all are in wireless-testing.
Ok, problem is reproduced with wireless-testing, too. I also have hangs
when unloading the driver sometimes (loaded with fwlps=0), but I wasn't
able to get a trace either.
>
> > (This problem was also reproduced with 3.2.X)
> >
> >
> >> to say that I cannot duplicate your results, and that we need to explore the
> >> rest of your system to see what is different.
> >>
> >> Is your system 32- or 64-bit? How many cores in your CPU? Have you tried
> > 32 bits, SoC is an AMD E1-1200, with 2 cores.
>
> My search shows that the clock rate for this processor is 1.4 GHz. The only
> system that I have that can run the RTL8188CE is 2 GHz.
>
> >> connecting to more than one AP? What is the make and model of the AP that needs
> > I've tried two different APs, with and without keys (open and WPA2)
> >
> >> to have power save turned off? What is the mode of operation?
> > One is a Linksys WRT54GL with dd-wrt(WPA2), the other(Open) is a
> > carrier's set top box/router (Freebox).
>
> I thought the locking in the Power Save routines was OK, but perhaps there is
> some raciness that is exposed by the difference in speeds. The lack of
> diagnostics on the netconsole is a problem.
>
> Do you have kernel debugging turned on for the lock diagnostics, and the Detect
> Hung Task configuration variable? Perhaps we can get some logging that way.
I enabled the following options:
CONFIG_LOCKDEP_SUPPORT=y
CONFIG_LOCKDEP=y
CONFIG_DEBUG_LOCKDEP=y
CONFIG_DEBUG_SPINLOCK=y
CONFIG_DEBUG_MUTEXES=y
But it didn't detect any lockup yet, and machine still hangs. I'll try
CONFIG_PROVE_LOCKING.
Enabling CONFIG_DEBUG_INFO seems to make the problem go away...
>
> Have you tried the vendor's driver from http://www.realtek.com? I don't think it will
> make a difference, but it could be worth a try.
Nope, not yet.
>
> Larry
(sorry for the delay, I got caught up in other things, and forgot to send
answer email)
On Wed, 16 May 2012 11:18:16 -0500, Larry Finger <[email protected]> wrote :
> On 05/16/2012 05:41 AM, Anisse Astier wrote:
> > I enabled the following options:
> > CONFIG_LOCKDEP_SUPPORT=y
> > CONFIG_LOCKDEP=y
> > CONFIG_DEBUG_LOCKDEP=y
> > CONFIG_DEBUG_SPINLOCK=y
> > CONFIG_DEBUG_MUTEXES=y
> > But it didn't detect any lockup yet, and machine still hangs. I'll try
> > CONFIG_PROVE_LOCKING.
> >
> > Enabling CONFIG_DEBUG_INFO seems to make the problem go away...
>
> That may explain why I never see the problem. I do not think I have ever
> generated a kernel with that parameter disabled, and my distro's kernels have it
> enabled. Does anything new get logged (dmesg output) when that parameter is enabled?
>
There's no zero difference in dmesg output. Have you tried with
debug_info disabled?
Unfortunately, I've lost access to this hardware, so I won't be able to
do more test for quite some time.
Hi,
On Thu, 24 May 2012 19:10:37 +0200, Anisse Astier <[email protected]> wrote :
> (sorry for the delay, I got caught up in other things, and forgot to send
> answer email)
>
> On Wed, 16 May 2012 11:18:16 -0500, Larry Finger <[email protected]> wrote :
>
> > On 05/16/2012 05:41 AM, Anisse Astier wrote:
> > > I enabled the following options:
> > > CONFIG_LOCKDEP_SUPPORT=y
> > > CONFIG_LOCKDEP=y
> > > CONFIG_DEBUG_LOCKDEP=y
> > > CONFIG_DEBUG_SPINLOCK=y
> > > CONFIG_DEBUG_MUTEXES=y
> > > But it didn't detect any lockup yet, and machine still hangs. I'll try
> > > CONFIG_PROVE_LOCKING.
> > >
> > > Enabling CONFIG_DEBUG_INFO seems to make the problem go away...
> >
> > That may explain why I never see the problem. I do not think I have ever
> > generated a kernel with that parameter disabled, and my distro's kernels have it
> > enabled. Does anything new get logged (dmesg output) when that parameter is enabled?
> >
>
> There's no zero difference in dmesg output. Have you tried with
> debug_info disabled?
>
> Unfortunately, I've lost access to this hardware, so I won't be able to
> do more test for quite some time.
I got hold of the hardware again. Problem is still here with today's
wireless-next (a4881cc45a3fab).
PROVE_LOCKING doesn't find any discrepancy.
I tried the new firmware files as well, no change.
Regards,
Anisse