Re: [PATCH 2/4] mips: cm: Add CM GCR and L2-sync base address getters declarations

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]


On Tue, Feb 20, 2024 at 06:24:25PM +0100, Thomas Bogendoerfer wrote:
> On Thu, Feb 15, 2024 at 08:17:27PM +0300, Serge Semin wrote:
> > Based on the design pattern utilized in the CM GCR and L2-sync base
> > address getters implementation the platform-specific code is capable to
> > re-define the getters and re-use the weakly defined initial versions. But
> > since the re-definition is supposed to be done in another source file the
> > interface methods have been globally defined which in its turn causes the
> > "no previous prototype" warning printed should the re-definition is
> > finally introduced. Since without the global declarations the pattern can
> > be considered as incomplete and causing the warning printed, fix it by
> > providing the respective methods prototype declarations in
> > "arch/mips/include/asm/mips-cm.h".
> > 
> > Signed-off-by: Serge Semin <fancer.lancer@xxxxxxxxx>
> > 
> > ---
> > 
> > Note as I mentioned in the previous patch, since the weak implementation
> > of the getters isn't utilized other than as a default implementation of
> > the original methods, we can convert the denoted pattern to a simple
> > __weak attributed methods. Let me know if that would be more preferable.

> how about simply remove __mips_cm_l2sync_phys_base() and do everything
> via mips_cm_phys_base(). And at the moment without anyone overriding
> mips_cm_phys_base I tend to keep static without __weak. If someone
> needs, we can change it. Does this make sense ?

To be honest my arch code (not submitted yet) do override the
mips_cm_l2sync_phys_base() method. The memory just behind the CM2
region is hardwired for the EEPROM mapping. So the default
__mips_cm_l2sync_phys_base() implementation selecting the L2-sync with
(mips_cm_phys_base() + MIPS_CM_GCR_SIZE) isn't suitable for my

Note what you suggest will require the CM and CM L2-sync probe code
logic redevelopment. The mips_cm_phys_base() method is currently
dedicated for a single purpose: return the CM GCR physical base address.
So is the mips_cm_l2sync_phys_base() method. Merging the later method
into the former one will cause the mips_cm_phys_base() function
looking less neat. We will also require to have the mips_cm_probe()
method keeping the CM L2-sync physical base addresses around and then
passing it to mips_cm_probe_l2sync() or having the later method
embedded into mips_cm_probe(). All of that won't look as good as the
current implementation.

What about instead of that I'll just convert both mips_cm_phys_base()
and mips_cm_l2sync_phys_base() to being defined with the underscored
methods body and assign the __weak attribute to them?


> Thomas.
> -- 
> Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
> good idea.                                                [ RFC1925, 2.3 ]

[Index of Archives]     [LKML Archive]     [Linux ARM Kernel]     [Linux ARM]     [Git]     [Yosemite News]     [Linux SCSI]     [Linux Hams]

  Powered by Linux