[U-Boot] [PATCH v1 1/1] cmd: adding malloc, math, and strcmp commands to u-boot

[Tom Rini]

On Wed, Nov 20, 2019 at 02:50:53PM -0800, Vladimir Olovyannikov wrote:
> Hi Tom,
> 
> > -----Original Message-----
> > From: Tom Rini [mailto:trini at konsulko.com]
> > Sent: Tuesday, November 19, 2019 12:38 PM
> > To: Vladimir Olovyannikov <vladimir.olovyannikov at broadcom.com>
> > Cc: u-boot at lists.denx.de; Suji Velupiallai
> <suji.velupillai at broadcom.com>;
> > Heinrich Schuchardt <xypron.glpk at gmx.de>
> > Subject: Re: [U-Boot] [PATCH v1 1/1] cmd: adding malloc, math, and
> strcmp
> > commands to u-boot
> >
> > On Mon, Nov 18, 2019 at 04:26:30PM -0800, Vladimir Olovyannikov wrote:
> >
> > > cmd: adding malloc, math, and strcmp u-boot commands.
> > > - malloc  supports allocation of heap memory and free allocated memory
> > >           via u-boot command line.
> >
> > Can you expand on how this is used in a script?  I'm not sure I see that
> > exactly.
> I am upstreaming the new bcm platform. Here is an excerpt from the script
> which uses
> these:
> #define SD_UPDATE \
> 	"sd_update="\
> 	"if malloc tmp 2000; then "\
> 	"else "\
> 		"echo [sd_update] malloc 2000 bytes ** FAILED **;"\
> 		"exit;"\
> 	"fi;"\
> 	"if fatload mmc ${sd_device_number} ${tmp} "\
> 	"${sd_update_prefix}.sd-update; then "\
> 	"else "\
> 		"echo [sd_update] fatload ${sd_update_prefix}.sd-update "\
> 		"** FAILED **;"\
> 		"exit;"\
> 	"fi;"\
> 	"if source ${tmp}; then "\
> 	"else "\
> 		"echo [sd_update] Executing script ** FAILED **;"\
> 		"exit;"\
> 	"fi;"\
> 	"if free tmp; then "\
> 	"else " \
> 		"echo [sd_update] free 2000 bytes ** FAILED **;"\
> 		"exit;"\
> 	"fi \0"
> ;
> 	"if math add filesize filesize 1FF; then "\
> 	"else "\
> 		"echo [mmc_flash_image_rsa] math add command ** FAILED
> **;"\
> 		"exit;"\
> 	"fi;"\
> 	"if math div fileblocks filesize 200; then "\
> 	"else "\
> 		"echo [mmc_flash_image_rsa] math div command ** FAILED
> **;"\
> 		"exit;"\
> 	"fi;"\

Ah, interesting.  So you're malloc'ing the memory location that you load
in to rather than the usual method of using known variables and
locations for the SoC.  I would discourage this as we generally do not
have a large malloc pool available.

> > Also:
> > > +config CMD_MALLOC
> > > +	bool "malloc"
> > > +	default y
> > > +	help
> > > +	  Supports allocation of heap memory and free allocated memory
> > commands.
> > > +	  These commands are used by u-boot scripts.
> > > +
> > > +config CMD_MATH
> > > +	bool "math"
> > > +	default y
> > > +	help
> > > +	  Provides math commands such as add, sub, mul, div, shift,
> > > +	  convert decimal to hex functionalities to be available in the
> script.
> >
> > First, why do we need this, rather than using setexpr ?
> I agree, the platform needs to use setexpr as it contains all math
> operations.
> >
> > > +
> > >  config CMD_SOURCE
> > >  	bool "source"
> > >  	default y
> > > @@ -1301,6 +1315,13 @@ config CMD_SETEXPR
> > >  	  Also supports loading the value at a memory location into a
> variable.
> > >  	  If CONFIG_REGEX is enabled, setexpr also supports a gsub
> function.
> > >
> > > +config CMD_STRCMP
> > > +	bool "strcmp"
> > > +	default y
> > > +	help
> > > +	  Provides string compare command feature to u-boot scripts.
> >
> > Second, new commands must not default to y, but they should be enabled
> > on sandbox and new test.py tests added for them.  Thanks!
> OK, thank you.
> Is there an example/doc on how to enable these on sandbox and to provide
> tests?

Unfortunately there is not (but would be greatly appreciated).  There
are a lot of examples under test/py/ however.  Thanks!

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20191120/c4bfcfbb/attachment.sig>