Re: v3.19: Nokia N900 (qemu) - omap_aes error

[Pali RohÃr]

On Wednesday 18 February 2015 22:36:41 Pali RohÃr wrote:
> On Monday 09 February 2015 12:25:00 Pali RohÃr wrote:
> > On Saturday 31 January 2015 16:29:36 Pali RohÃr wrote:
> > > Hello,
> > > 
> > > when I enable problematic omap_aes.ko driver (also in DT)
> > > for Nokia N900 machine in qemu (it emulates OMAP GP
> > > device, not HS) I get this error message in dmesg at
> > > modprobe time:
> > > 
> > > [   37.660339] ------------[ cut here ]------------
> > > [   37.663940] WARNING: CPU: 0 PID: 173 at arch/arm/mach-
> > > omap2/omap_hwmod.c:2182 _idle+0x28/0x64()
> > > [   37.668762] omap_hwmod: aes: idle state can only be
> > > entered from enabled state
> > > [   37.672271] Modules linked in: omap_aes(+)
> > > [   37.675933] CPU: 0 PID: 173 Comm: modprobe Not tainted
> > > 3.19.0- rc5+ #299
> > > [   37.679534] Hardware name: Nokia RX-51 board
> > > [   37.683227] [<c0012678>] (unwind_backtrace) from
> > > [<c0010d44>] (show_stack+0x10/0x14)
> > > [   37.686920] [<c0010d44>] (show_stack) from [<c0032194>]
> > > (warn_slowpath_common+0x84/0xac)
> > > [   37.690765] [<c0032194>] (warn_slowpath_common) from
> > > [<c003223c>] (warn_slowpath_fmt+0x2c/0x3c)
> > > [   37.694366] [<c003223c>] (warn_slowpath_fmt) from
> > > [<c0020cb0>] (_idle+0x28/0x64)
> > > [   37.698791] [<c0020cb0>] (_idle) from [<c0021078>]
> > > (omap_hwmod_idle+0x30/0x78)
> > > [   37.702636] [<c0021078>] (omap_hwmod_idle) from
> > > [<c00222c4>] (omap_device_idle+0x48/0x6c)
> > > [   37.707153] [<c00222c4>] (omap_device_idle) from
> > > [<c0022304>] (_od_runtime_suspend+0x1c/0x24)
> > > [   37.710723] [<c0022304>] (_od_runtime_suspend) from
> > > [<c0269680>] (__rpm_callback+0x8c/0xdc)
> > > [   37.714385] [<c0269680>] (__rpm_callback) from
> > > [<c0269740>] (rpm_callback+0x70/0x88)
> > > [   37.717987] [<c0269740>] (rpm_callback) from
> > > [<c0269f74>] (rpm_suspend+0x2e8/0x5a4)
> > > [   37.721710] [<c0269f74>] (rpm_suspend) from
> > > [<c026a6ec>] (__pm_runtime_idle+0x58/0x9c)
> > > [   37.725402] [<c026a6ec>] (__pm_runtime_idle) from
> > > [<bf00129c>] (omap_aes_probe+0x15c/0x448 [omap_aes])
> > > [   37.733489] [<bf00129c>] (omap_aes_probe [omap_aes])
> > > from [<c0263d78>] (platform_drv_probe+0x48/0x90)
> > > [   37.740936] [<c0263d78>] (platform_drv_probe) from
> > > [<c02624ec>] (really_probe+0xac/0x1e0)
> > > [   37.745025] [<c02624ec>] (really_probe) from
> > > [<c0262718>] (driver_probe_device+0x30/0x48)
> > > [   37.749847] [<c0262718>] (driver_probe_device) from
> > > [<c0262790>] (__driver_attach+0x60/0x84)
> > > [   37.753601] [<c0262790>] (__driver_attach) from
> > > [<c0261070>] (bus_for_each_dev+0x50/0x84)
> > > [   37.758422] [<c0261070>] (bus_for_each_dev) from
> > > [<c0261e20>] (bus_add_driver+0xac/0x1bc)
> > > [   37.762207] [<c0261e20>] (bus_add_driver) from
> > > [<c02630c8>] (driver_register+0x9c/0xe0)
> > > [   37.765899] [<c02630c8>] (driver_register) from
> > > [<c0008798>] (do_one_initcall+0x100/0x1b0)
> > > [   37.769653] [<c0008798>] (do_one_initcall) from
> > > [<c0085ca4>] (do_init_module+0x34/0x158)
> > > [   37.773590] [<c0085ca4>] (do_init_module) from
> > > [<c00868d8>] (SyS_init_module+0x54/0x64)
> > > [   37.777252] [<c00868d8>] (SyS_init_module) from
> > > [<c000dbe0>] (ret_fast_syscall+0x0/0x30)
> > > [   37.780975] ---[ end trace 86c3e025b92e700d ]---
> > > [   37.784667] omap-aes 480c5000.aes: OMAP AES hw accel
> > > rev: 0.0 [   37.789794] of_dma_request_slave_channel:
> > > dma-names property of node '/ocp/aes@480c5000' missing or
> > > empty [   37.797668] omap-aes 480c5000.aes: Unable to
> > > request in DMA channel
> > > [   37.801391]    omap_aes_dma_init: error: -12
> > > 
> > > Error is not fatal, kernel does not crash and qemu
> > > emulation working fine. But error is there.
> > > 
> > > I'm sure that qemu does not support omap aes hw emulation
> > > and so I except error message about missing hw part (or
> > > something else). But not above unwind_backtrace error
> > > message.
> > > 
> > > Any idea why it happens? It is probably not related to AES
> > > but to DMA. It is for OMAP GP device (but emulated), so
> > > can TI comment this problem (as this is not HS under
> > > NDA)?
> > 
> > Hello,
> > 
> > what does that error line means?
> > 
> > [   37.668762] omap_hwmod: aes: idle state can only be
> > entered from enabled state
> > 
> > Is not it bug in omap_hwmod.c code?
> 
> In qemu dmesg log I see this debug line:
> 
> [    0.297851] omap_hwmod: aes: cannot be enabled for reset
> (3)
> 
> So aes is in disabled state.
> 
> Should not omap-aes driver check for aes state if is enabled?

Above dmesg error should be fixed by my patch sent to ML:

ARM: OMAP2+: Return correct error values from device and hwmod

-- 
Pali RohÃr
pali.rohar@xxxxxxxxx

Attachment: signature.asc
Description: This is a digitally signed message part.