From: dbrownell Date: Thu, 17 Sep 2009 07:56:24 +0000 (+0000) Subject: Doc update: mention how ARM's WFI instruction affects X-Git-Tag: v0.3.0-rc0~233 X-Git-Url: https://git.sur5r.net/?a=commitdiff_plain;h=e18bd3b55e3351a38f6bcc5ec17373e022210e9e;p=openocd Doc update: mention how ARM's WFI instruction affects JTAG clocking by gating the core clock, and workarounds. Most details are with the "halt" command, which is one of the first places this issue will be noticed. git-svn-id: svn://svn.berlios.de/openocd/trunk@2718 b42882b7-edfa-0310-969c-e2dbd0fdcd60 --- diff --git a/doc/openocd.texi b/doc/openocd.texi index 013d08a1..9960468c 100644 --- a/doc/openocd.texi +++ b/doc/openocd.texi @@ -1055,7 +1055,19 @@ access uses the CPU or to prevent conflicting CPU access. Before your @code{reset-init} handler has set up the PLLs and clocking, you may need to use a low JTAG clock rate; then you'd increase it later. -(The rule of thumb for ARM-based processors is 1/8 the CPU clock.) +For most ARM-based processors the fastest JTAG clock@footnote{A FAQ +@uref{http://www.arm.com/support/faqdev/4170.html} gives details.} +is one sixth of the CPU clock; or one eighth for ARM11 cores. +Consult chip documentation to determine the peak JTAG clock rate, +which might be less than that. + +@quotation Warning +On most ARMs, JTAG clock detection is coupled to the core clock, so +software using a @option{wait for interrupt} operation blocks JTAG access. +Adaptive clocking provides a partial workaround, but a more complete +solution just avoids using that instruction with JTAG debuggers. +@end quotation + If the board supports adaptive clocking, use the @command{jtag_rclk} command, in case your board is used with JTAG adapter which also supports it. Otherwise use @command{jtag_khz}. @@ -1785,9 +1797,10 @@ JTAG interfaces usually support a limited number of speeds. The speed actually used won't be faster than the speed specified. -As a rule of thumb, if you specify a clock rate make -sure the JTAG clock is no more than @math{1/6th CPU-Clock}. -This is especially true for synthesized cores (ARMxxx-S). +Chip data sheets generally include a top JTAG clock rate. +The actual rate is often a function of a CPU core clock, +and is normally less than that peak rate. +For example, most ARM cores accept at most one sixth of the CPU clock. Speed 0 (khz) selects RTCK method. @xref{FAQ RTCK}. @@ -1799,6 +1812,7 @@ support it, an error is returned when you try to use RTCK. @end deffn @defun jtag_rclk fallback_speed_kHz +@cindex adaptive clocking @cindex RTCK This Tcl proc (defined in @file{startup.tcl}) attempts to enable RTCK/RCLK. If that fails (maybe the interface, board, or target doesn't @@ -4321,6 +4335,31 @@ Otherwise these behave the same: wait up to @var{ms} milliseconds, or 5 seconds if there is no parameter, for the target to halt (and enter debug mode). Using 0 as the @var{ms} parameter prevents OpenOCD from waiting. + +@quotation Warning +On ARM cores, software using the @emph{wait for interrupt} operation +often blocks the JTAG access needed by a @command{halt} command. +This is because that operation also puts the core into a low +power mode by gating the core clock; +but the core clock is needed to detect JTAG clock transitions. + +One partial workaround uses adaptive clocking: when the core is +interrupted the operation completes, then JTAG clocks are accepted +at least until the interrupt handler completes. +However, this workaround is often unusable since the processor, board, +and JTAG adapter must all support adaptive JTAG clocking. +Also, it can't work until an interrupt is issued. + +A more complete workaround is to not use that operation while you +work with a JTAG debugger. +Tasking environments generaly have idle loops where the body is the +@emph{wait for interrupt} operation. +(On older cores, it is a coprocessor action; +newer cores have a @option{wfi} instruction.) +Such loops can just remove that operation, at the cost of higher +power consumption (because the CPU is needlessly clocked). +@end quotation + @end deffn @deffn Command resume [address] @@ -5984,9 +6023,10 @@ Often this is a perfectly acceptable solution. In most simple terms: Often the JTAG clock must be 1/10 to 1/12 of the target clock speed. But what that ``magic division'' is varies -depending on the chips on your board. @b{ARM rule of thumb} Most ARM -based systems require an 8:1 division. @b{Xilinx rule of thumb} is -1/12 the clock speed. +depending on the chips on your board. +@b{ARM rule of thumb} Most ARM based systems require an 6:1 division; +ARM11 cores use an 8:1 division. +@b{Xilinx rule of thumb} is 1/12 the clock speed. Note: Many FTDI2232C based JTAG dongles are limited to 6MHz. @@ -5999,11 +6039,18 @@ have a special debug mode in your application that does a ``high power sleep''. If you are careful - 98% of your problems can be debugged this way. +Note that on ARM you may need to avoid using the @emph{wait for interrupt} +operation in your idle loops even if you don't otherwise change the CPU +clock rate. +That operation gates the CPU clock, and thus the JTAG clock; which +prevents JTAG access. One consequence is not being able to @command{halt} +cores which are executing that @emph{wait for interrupt} operation. + To set the JTAG frequency use the command: @example - # Example: 1.234MHz - jtag_khz 1234 +# Example: 1.234MHz +jtag_khz 1234 @end example