<html lang="en">
<head>
<title>SH Options - Using the GNU Compiler Collection (GCC)</title>
<meta http-equiv="Content-Type" content="text/html">
<meta name="description" content="Using the GNU Compiler Collection (GCC)">
<meta name="generator" content="makeinfo 4.13">
<link title="Top" rel="start" href="index.html#Top">
<link rel="up" href="Submodel-Options.html#Submodel-Options" title="Submodel Options">
<link rel="prev" href="Score-Options.html#Score-Options" title="Score Options">
<link rel="next" href="Solaris-2-Options.html#Solaris-2-Options" title="Solaris 2 Options">
<link href="http://www.gnu.org/software/texinfo/" rel="generator-home" title="Texinfo Homepage">
<!--
Copyright (C) 1988-2015 Free Software Foundation, Inc.
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License, Version 1.3 or
any later version published by the Free Software Foundation; with the
Invariant Sections being ``Funding Free Software'', the Front-Cover
Texts being (a) (see below), and with the Back-Cover Texts being (b)
(see below). A copy of the license is included in the section entitled
``GNU Free Documentation License''.
(a) The FSF's Front-Cover Text is:
A GNU Manual
(b) The FSF's Back-Cover Text is:
You have freedom to copy and modify this GNU Manual, like GNU
software. Copies published by the Free Software Foundation raise
funds for GNU development.-->
<meta http-equiv="Content-Style-Type" content="text/css">
<style type="text/css"><!--
pre.display { font-family:inherit }
pre.format { font-family:inherit }
pre.smalldisplay { font-family:inherit; font-size:smaller }
pre.smallformat { font-family:inherit; font-size:smaller }
pre.smallexample { font-size:smaller }
pre.smalllisp { font-size:smaller }
span.sc { font-variant:small-caps }
span.roman { font-family:serif; font-weight:normal; }
span.sansserif { font-family:sans-serif; font-weight:normal; }
--></style>
</head>
<body>
<div class="node">
<a name="SH-Options"></a>
<p>
Next: <a rel="next" accesskey="n" href="Solaris-2-Options.html#Solaris-2-Options">Solaris 2 Options</a>,
Previous: <a rel="previous" accesskey="p" href="Score-Options.html#Score-Options">Score Options</a>,
Up: <a rel="up" accesskey="u" href="Submodel-Options.html#Submodel-Options">Submodel Options</a>
<hr>
</div>
<h4 class="subsection">3.17.41 SH Options</h4>
<p>These ‘<samp><span class="samp">-m</span></samp>’ options are defined for the SH implementations:
<dl>
<dt><code>-m1</code><dd><a name="index-m1-2433"></a>Generate code for the SH1.
<br><dt><code>-m2</code><dd><a name="index-m2-2434"></a>Generate code for the SH2.
<br><dt><code>-m2e</code><dd>Generate code for the SH2e.
<br><dt><code>-m2a-nofpu</code><dd><a name="index-m2a_002dnofpu-2435"></a>Generate code for the SH2a without FPU, or for a SH2a-FPU in such a way
that the floating-point unit is not used.
<br><dt><code>-m2a-single-only</code><dd><a name="index-m2a_002dsingle_002donly-2436"></a>Generate code for the SH2a-FPU, in such a way that no double-precision
floating-point operations are used.
<br><dt><code>-m2a-single</code><dd><a name="index-m2a_002dsingle-2437"></a>Generate code for the SH2a-FPU assuming the floating-point unit is in
single-precision mode by default.
<br><dt><code>-m2a</code><dd><a name="index-m2a-2438"></a>Generate code for the SH2a-FPU assuming the floating-point unit is in
double-precision mode by default.
<br><dt><code>-m3</code><dd><a name="index-m3-2439"></a>Generate code for the SH3.
<br><dt><code>-m3e</code><dd><a name="index-m3e-2440"></a>Generate code for the SH3e.
<br><dt><code>-m4-nofpu</code><dd><a name="index-m4_002dnofpu-2441"></a>Generate code for the SH4 without a floating-point unit.
<br><dt><code>-m4-single-only</code><dd><a name="index-m4_002dsingle_002donly-2442"></a>Generate code for the SH4 with a floating-point unit that only
supports single-precision arithmetic.
<br><dt><code>-m4-single</code><dd><a name="index-m4_002dsingle-2443"></a>Generate code for the SH4 assuming the floating-point unit is in
single-precision mode by default.
<br><dt><code>-m4</code><dd><a name="index-m4-2444"></a>Generate code for the SH4.
<br><dt><code>-m4-100</code><dd><a name="index-m4_002d100-2445"></a>Generate code for SH4-100.
<br><dt><code>-m4-100-nofpu</code><dd><a name="index-m4_002d100_002dnofpu-2446"></a>Generate code for SH4-100 in such a way that the
floating-point unit is not used.
<br><dt><code>-m4-100-single</code><dd><a name="index-m4_002d100_002dsingle-2447"></a>Generate code for SH4-100 assuming the floating-point unit is in
single-precision mode by default.
<br><dt><code>-m4-100-single-only</code><dd><a name="index-m4_002d100_002dsingle_002donly-2448"></a>Generate code for SH4-100 in such a way that no double-precision
floating-point operations are used.
<br><dt><code>-m4-200</code><dd><a name="index-m4_002d200-2449"></a>Generate code for SH4-200.
<br><dt><code>-m4-200-nofpu</code><dd><a name="index-m4_002d200_002dnofpu-2450"></a>Generate code for SH4-200 without in such a way that the
floating-point unit is not used.
<br><dt><code>-m4-200-single</code><dd><a name="index-m4_002d200_002dsingle-2451"></a>Generate code for SH4-200 assuming the floating-point unit is in
single-precision mode by default.
<br><dt><code>-m4-200-single-only</code><dd><a name="index-m4_002d200_002dsingle_002donly-2452"></a>Generate code for SH4-200 in such a way that no double-precision
floating-point operations are used.
<br><dt><code>-m4-300</code><dd><a name="index-m4_002d300-2453"></a>Generate code for SH4-300.
<br><dt><code>-m4-300-nofpu</code><dd><a name="index-m4_002d300_002dnofpu-2454"></a>Generate code for SH4-300 without in such a way that the
floating-point unit is not used.
<br><dt><code>-m4-300-single</code><dd><a name="index-m4_002d300_002dsingle-2455"></a>Generate code for SH4-300 in such a way that no double-precision
floating-point operations are used.
<br><dt><code>-m4-300-single-only</code><dd><a name="index-m4_002d300_002dsingle_002donly-2456"></a>Generate code for SH4-300 in such a way that no double-precision
floating-point operations are used.
<br><dt><code>-m4-340</code><dd><a name="index-m4_002d340-2457"></a>Generate code for SH4-340 (no MMU, no FPU).
<br><dt><code>-m4-500</code><dd><a name="index-m4_002d500-2458"></a>Generate code for SH4-500 (no FPU). Passes <samp><span class="option">-isa=sh4-nofpu</span></samp> to the
assembler.
<br><dt><code>-m4a-nofpu</code><dd><a name="index-m4a_002dnofpu-2459"></a>Generate code for the SH4al-dsp, or for a SH4a in such a way that the
floating-point unit is not used.
<br><dt><code>-m4a-single-only</code><dd><a name="index-m4a_002dsingle_002donly-2460"></a>Generate code for the SH4a, in such a way that no double-precision
floating-point operations are used.
<br><dt><code>-m4a-single</code><dd><a name="index-m4a_002dsingle-2461"></a>Generate code for the SH4a assuming the floating-point unit is in
single-precision mode by default.
<br><dt><code>-m4a</code><dd><a name="index-m4a-2462"></a>Generate code for the SH4a.
<br><dt><code>-m4al</code><dd><a name="index-m4al-2463"></a>Same as <samp><span class="option">-m4a-nofpu</span></samp>, except that it implicitly passes
<samp><span class="option">-dsp</span></samp> to the assembler. GCC doesn't generate any DSP
instructions at the moment.
<br><dt><code>-m5-32media</code><dd><a name="index-m5_002d32media-2464"></a>Generate 32-bit code for SHmedia.
<br><dt><code>-m5-32media-nofpu</code><dd><a name="index-m5_002d32media_002dnofpu-2465"></a>Generate 32-bit code for SHmedia in such a way that the
floating-point unit is not used.
<br><dt><code>-m5-64media</code><dd><a name="index-m5_002d64media-2466"></a>Generate 64-bit code for SHmedia.
<br><dt><code>-m5-64media-nofpu</code><dd><a name="index-m5_002d64media_002dnofpu-2467"></a>Generate 64-bit code for SHmedia in such a way that the
floating-point unit is not used.
<br><dt><code>-m5-compact</code><dd><a name="index-m5_002dcompact-2468"></a>Generate code for SHcompact.
<br><dt><code>-m5-compact-nofpu</code><dd><a name="index-m5_002dcompact_002dnofpu-2469"></a>Generate code for SHcompact in such a way that the
floating-point unit is not used.
<br><dt><code>-mb</code><dd><a name="index-mb-2470"></a>Compile code for the processor in big-endian mode.
<br><dt><code>-ml</code><dd><a name="index-ml-2471"></a>Compile code for the processor in little-endian mode.
<br><dt><code>-mdalign</code><dd><a name="index-mdalign-2472"></a>Align doubles at 64-bit boundaries. Note that this changes the calling
conventions, and thus some functions from the standard C library do
not work unless you recompile it first with <samp><span class="option">-mdalign</span></samp>.
<br><dt><code>-mrelax</code><dd><a name="index-mrelax-2473"></a>Shorten some address references at link time, when possible; uses the
linker option <samp><span class="option">-relax</span></samp>.
<br><dt><code>-mbigtable</code><dd><a name="index-mbigtable-2474"></a>Use 32-bit offsets in <code>switch</code> tables. The default is to use
16-bit offsets.
<br><dt><code>-mbitops</code><dd><a name="index-mbitops-2475"></a>Enable the use of bit manipulation instructions on SH2A.
<br><dt><code>-mfmovd</code><dd><a name="index-mfmovd-2476"></a>Enable the use of the instruction <code>fmovd</code>. Check <samp><span class="option">-mdalign</span></samp> for
alignment constraints.
<br><dt><code>-mrenesas</code><dd><a name="index-mrenesas-2477"></a>Comply with the calling conventions defined by Renesas.
<br><dt><code>-mno-renesas</code><dd><a name="index-mno_002drenesas-2478"></a>Comply with the calling conventions defined for GCC before the Renesas
conventions were available. This option is the default for all
targets of the SH toolchain.
<br><dt><code>-mnomacsave</code><dd><a name="index-mnomacsave-2479"></a>Mark the <code>MAC</code> register as call-clobbered, even if
<samp><span class="option">-mrenesas</span></samp> is given.
<br><dt><code>-mieee</code><dt><code>-mno-ieee</code><dd><a name="index-mieee-2480"></a><a name="index-mno_002dieee-2481"></a>Control the IEEE compliance of floating-point comparisons, which affects the
handling of cases where the result of a comparison is unordered. By default
<samp><span class="option">-mieee</span></samp> is implicitly enabled. If <samp><span class="option">-ffinite-math-only</span></samp> is
enabled <samp><span class="option">-mno-ieee</span></samp> is implicitly set, which results in faster
floating-point greater-equal and less-equal comparisons. The implcit settings
can be overridden by specifying either <samp><span class="option">-mieee</span></samp> or <samp><span class="option">-mno-ieee</span></samp>.
<br><dt><code>-minline-ic_invalidate</code><dd><a name="index-minline_002dic_005finvalidate-2482"></a>Inline code to invalidate instruction cache entries after setting up
nested function trampolines.
This option has no effect if <samp><span class="option">-musermode</span></samp> is in effect and the selected
code generation option (e.g. <samp><span class="option">-m4</span></samp>) does not allow the use of the <code>icbi</code>
instruction.
If the selected code generation option does not allow the use of the <code>icbi</code>
instruction, and <samp><span class="option">-musermode</span></samp> is not in effect, the inlined code
manipulates the instruction cache address array directly with an associative
write. This not only requires privileged mode at run time, but it also
fails if the cache line had been mapped via the TLB and has become unmapped.
<br><dt><code>-misize</code><dd><a name="index-misize-2483"></a>Dump instruction size and location in the assembly code.
<br><dt><code>-mpadstruct</code><dd><a name="index-mpadstruct-2484"></a>This option is deprecated. It pads structures to multiple of 4 bytes,
which is incompatible with the SH ABI.
<br><dt><code>-matomic-model=</code><var>model</var><dd><a name="index-matomic_002dmodel_003d_0040var_007bmodel_007d-2485"></a>Sets the model of atomic operations and additional parameters as a comma
separated list. For details on the atomic built-in functions see
<a href="_005f_005fatomic-Builtins.html#g_t_005f_005fatomic-Builtins">__atomic Builtins</a>. The following models and parameters are supported:
<dl>
<dt>‘<samp><span class="samp">none</span></samp>’<dd>Disable compiler generated atomic sequences and emit library calls for atomic
operations. This is the default if the target is not <code>sh*-*-linux*</code>.
<br><dt>‘<samp><span class="samp">soft-gusa</span></samp>’<dd>Generate GNU/Linux compatible gUSA software atomic sequences for the atomic
built-in functions. The generated atomic sequences require additional support
from the interrupt/exception handling code of the system and are only suitable
for SH3* and SH4* single-core systems. This option is enabled by default when
the target is <code>sh*-*-linux*</code> and SH3* or SH4*. When the target is SH4A,
this option also partially utilizes the hardware atomic instructions
<code>movli.l</code> and <code>movco.l</code> to create more efficient code, unless
‘<samp><span class="samp">strict</span></samp>’ is specified.
<br><dt>‘<samp><span class="samp">soft-tcb</span></samp>’<dd>Generate software atomic sequences that use a variable in the thread control
block. This is a variation of the gUSA sequences which can also be used on
SH1* and SH2* targets. The generated atomic sequences require additional
support from the interrupt/exception handling code of the system and are only
suitable for single-core systems. When using this model, the ‘<samp><span class="samp">gbr-offset=</span></samp>’
parameter has to be specified as well.
<br><dt>‘<samp><span class="samp">soft-imask</span></samp>’<dd>Generate software atomic sequences that temporarily disable interrupts by
setting <code>SR.IMASK = 1111</code>. This model works only when the program runs
in privileged mode and is only suitable for single-core systems. Additional
support from the interrupt/exception handling code of the system is not
required. This model is enabled by default when the target is
<code>sh*-*-linux*</code> and SH1* or SH2*.
<br><dt>‘<samp><span class="samp">hard-llcs</span></samp>’<dd>Generate hardware atomic sequences using the <code>movli.l</code> and <code>movco.l</code>
instructions only. This is only available on SH4A and is suitable for
multi-core systems. Since the hardware instructions support only 32 bit atomic
variables access to 8 or 16 bit variables is emulated with 32 bit accesses.
Code compiled with this option is also compatible with other software
atomic model interrupt/exception handling systems if executed on an SH4A
system. Additional support from the interrupt/exception handling code of the
system is not required for this model.
<br><dt>‘<samp><span class="samp">gbr-offset=</span></samp>’<dd>This parameter specifies the offset in bytes of the variable in the thread
control block structure that should be used by the generated atomic sequences
when the ‘<samp><span class="samp">soft-tcb</span></samp>’ model has been selected. For other models this
parameter is ignored. The specified value must be an integer multiple of four
and in the range 0-1020.
<br><dt>‘<samp><span class="samp">strict</span></samp>’<dd>This parameter prevents mixed usage of multiple atomic models, even if they
are compatible, and makes the compiler generate atomic sequences of the
specified model only.
</dl>
<br><dt><code>-mtas</code><dd><a name="index-mtas-2486"></a>Generate the <code>tas.b</code> opcode for <code>__atomic_test_and_set</code>.
Notice that depending on the particular hardware and software configuration
this can degrade overall performance due to the operand cache line flushes
that are implied by the <code>tas.b</code> instruction. On multi-core SH4A
processors the <code>tas.b</code> instruction must be used with caution since it
can result in data corruption for certain cache configurations.
<br><dt><code>-mprefergot</code><dd><a name="index-mprefergot-2487"></a>When generating position-independent code, emit function calls using
the Global Offset Table instead of the Procedure Linkage Table.
<br><dt><code>-musermode</code><dt><code>-mno-usermode</code><dd><a name="index-musermode-2488"></a><a name="index-mno_002dusermode-2489"></a>Don't allow (allow) the compiler generating privileged mode code. Specifying
<samp><span class="option">-musermode</span></samp> also implies <samp><span class="option">-mno-inline-ic_invalidate</span></samp> if the
inlined code would not work in user mode. <samp><span class="option">-musermode</span></samp> is the default
when the target is <code>sh*-*-linux*</code>. If the target is SH1* or SH2*
<samp><span class="option">-musermode</span></samp> has no effect, since there is no user mode.
<br><dt><code>-multcost=</code><var>number</var><dd><a name="index-multcost_003d_0040var_007bnumber_007d-2490"></a>Set the cost to assume for a multiply insn.
<br><dt><code>-mdiv=</code><var>strategy</var><dd><a name="index-mdiv_003d_0040var_007bstrategy_007d-2491"></a>Set the division strategy to be used for integer division operations.
For SHmedia <var>strategy</var> can be one of:
<dl>
<dt>‘<samp><span class="samp">fp</span></samp>’<dd>Performs the operation in floating point. This has a very high latency,
but needs only a few instructions, so it might be a good choice if
your code has enough easily-exploitable ILP to allow the compiler to
schedule the floating-point instructions together with other instructions.
Division by zero causes a floating-point exception.
<br><dt>‘<samp><span class="samp">inv</span></samp>’<dd>Uses integer operations to calculate the inverse of the divisor,
and then multiplies the dividend with the inverse. This strategy allows
CSE and hoisting of the inverse calculation. Division by zero calculates
an unspecified result, but does not trap.
<br><dt>‘<samp><span class="samp">inv:minlat</span></samp>’<dd>A variant of ‘<samp><span class="samp">inv</span></samp>’ where, if no CSE or hoisting opportunities
have been found, or if the entire operation has been hoisted to the same
place, the last stages of the inverse calculation are intertwined with the
final multiply to reduce the overall latency, at the expense of using a few
more instructions, and thus offering fewer scheduling opportunities with
other code.
<br><dt>‘<samp><span class="samp">call</span></samp>’<dd>Calls a library function that usually implements the ‘<samp><span class="samp">inv:minlat</span></samp>’
strategy.
This gives high code density for <code>m5-*media-nofpu</code> compilations.
<br><dt>‘<samp><span class="samp">call2</span></samp>’<dd>Uses a different entry point of the same library function, where it
assumes that a pointer to a lookup table has already been set up, which
exposes the pointer load to CSE and code hoisting optimizations.
<br><dt>‘<samp><span class="samp">inv:call</span></samp>’<dt>‘<samp><span class="samp">inv:call2</span></samp>’<dt>‘<samp><span class="samp">inv:fp</span></samp>’<dd>Use the ‘<samp><span class="samp">inv</span></samp>’ algorithm for initial
code generation, but if the code stays unoptimized, revert to the ‘<samp><span class="samp">call</span></samp>’,
‘<samp><span class="samp">call2</span></samp>’, or ‘<samp><span class="samp">fp</span></samp>’ strategies, respectively. Note that the
potentially-trapping side effect of division by zero is carried by a
separate instruction, so it is possible that all the integer instructions
are hoisted out, but the marker for the side effect stays where it is.
A recombination to floating-point operations or a call is not possible
in that case.
<br><dt>‘<samp><span class="samp">inv20u</span></samp>’<dt>‘<samp><span class="samp">inv20l</span></samp>’<dd>Variants of the ‘<samp><span class="samp">inv:minlat</span></samp>’ strategy. In the case
that the inverse calculation is not separated from the multiply, they speed
up division where the dividend fits into 20 bits (plus sign where applicable)
by inserting a test to skip a number of operations in this case; this test
slows down the case of larger dividends. ‘<samp><span class="samp">inv20u</span></samp>’ assumes the case of a such
a small dividend to be unlikely, and ‘<samp><span class="samp">inv20l</span></samp>’ assumes it to be likely.
</dl>
<p>For targets other than SHmedia <var>strategy</var> can be one of:
<dl>
<dt>‘<samp><span class="samp">call-div1</span></samp>’<dd>Calls a library function that uses the single-step division instruction
<code>div1</code> to perform the operation. Division by zero calculates an
unspecified result and does not trap. This is the default except for SH4,
SH2A and SHcompact.
<br><dt>‘<samp><span class="samp">call-fp</span></samp>’<dd>Calls a library function that performs the operation in double precision
floating point. Division by zero causes a floating-point exception. This is
the default for SHcompact with FPU. Specifying this for targets that do not
have a double precision FPU defaults to <code>call-div1</code>.
<br><dt>‘<samp><span class="samp">call-table</span></samp>’<dd>Calls a library function that uses a lookup table for small divisors and
the <code>div1</code> instruction with case distinction for larger divisors. Division
by zero calculates an unspecified result and does not trap. This is the default
for SH4. Specifying this for targets that do not have dynamic shift
instructions defaults to <code>call-div1</code>.
</dl>
<p>When a division strategy has not been specified the default strategy is
selected based on the current target. For SH2A the default strategy is to
use the <code>divs</code> and <code>divu</code> instructions instead of library function
calls.
<br><dt><code>-maccumulate-outgoing-args</code><dd><a name="index-maccumulate_002doutgoing_002dargs-2492"></a>Reserve space once for outgoing arguments in the function prologue rather
than around each call. Generally beneficial for performance and size. Also
needed for unwinding to avoid changing the stack frame around conditional code.
<br><dt><code>-mdivsi3_libfunc=</code><var>name</var><dd><a name="index-mdivsi3_005flibfunc_003d_0040var_007bname_007d-2493"></a>Set the name of the library function used for 32-bit signed division to
<var>name</var>.
This only affects the name used in the ‘<samp><span class="samp">call</span></samp>’ and ‘<samp><span class="samp">inv:call</span></samp>’
division strategies, and the compiler still expects the same
sets of input/output/clobbered registers as if this option were not present.
<br><dt><code>-mfixed-range=</code><var>register-range</var><dd><a name="index-mfixed_002drange-2494"></a>Generate code treating the given register range as fixed registers.
A fixed register is one that the register allocator can not use. This is
useful when compiling kernel code. A register range is specified as
two registers separated by a dash. Multiple register ranges can be
specified separated by a comma.
<br><dt><code>-mindexed-addressing</code><dd><a name="index-mindexed_002daddressing-2495"></a>Enable the use of the indexed addressing mode for SHmedia32/SHcompact.
This is only safe if the hardware and/or OS implement 32-bit wrap-around
semantics for the indexed addressing mode. The architecture allows the
implementation of processors with 64-bit MMU, which the OS could use to
get 32-bit addressing, but since no current hardware implementation supports
this or any other way to make the indexed addressing mode safe to use in
the 32-bit ABI, the default is <samp><span class="option">-mno-indexed-addressing</span></samp>.
<br><dt><code>-mgettrcost=</code><var>number</var><dd><a name="index-mgettrcost_003d_0040var_007bnumber_007d-2496"></a>Set the cost assumed for the <code>gettr</code> instruction to <var>number</var>.
The default is 2 if <samp><span class="option">-mpt-fixed</span></samp> is in effect, 100 otherwise.
<br><dt><code>-mpt-fixed</code><dd><a name="index-mpt_002dfixed-2497"></a>Assume <code>pt*</code> instructions won't trap. This generally generates
better-scheduled code, but is unsafe on current hardware.
The current architecture
definition says that <code>ptabs</code> and <code>ptrel</code> trap when the target
anded with 3 is 3.
This has the unintentional effect of making it unsafe to schedule these
instructions before a branch, or hoist them out of a loop. For example,
<code>__do_global_ctors</code>, a part of <samp><span class="file">libgcc</span></samp>
that runs constructors at program
startup, calls functions in a list which is delimited by −1. With the
<samp><span class="option">-mpt-fixed</span></samp> option, the <code>ptabs</code> is done before testing against −1.
That means that all the constructors run a bit more quickly, but when
the loop comes to the end of the list, the program crashes because <code>ptabs</code>
loads −1 into a target register.
<p>Since this option is unsafe for any
hardware implementing the current architecture specification, the default
is <samp><span class="option">-mno-pt-fixed</span></samp>. Unless specified explicitly with
<samp><span class="option">-mgettrcost</span></samp>, <samp><span class="option">-mno-pt-fixed</span></samp> also implies <samp><span class="option">-mgettrcost=100</span></samp>;
this deters register allocation from using target registers for storing
ordinary integers.
<br><dt><code>-minvalid-symbols</code><dd><a name="index-minvalid_002dsymbols-2498"></a>Assume symbols might be invalid. Ordinary function symbols generated by
the compiler are always valid to load with
<code>movi</code>/<code>shori</code>/<code>ptabs</code> or
<code>movi</code>/<code>shori</code>/<code>ptrel</code>,
but with assembler and/or linker tricks it is possible
to generate symbols that cause <code>ptabs</code> or <code>ptrel</code> to trap.
This option is only meaningful when <samp><span class="option">-mno-pt-fixed</span></samp> is in effect.
It prevents cross-basic-block CSE, hoisting and most scheduling
of symbol loads. The default is <samp><span class="option">-mno-invalid-symbols</span></samp>.
<br><dt><code>-mbranch-cost=</code><var>num</var><dd><a name="index-mbranch_002dcost_003d_0040var_007bnum_007d-2499"></a>Assume <var>num</var> to be the cost for a branch instruction. Higher numbers
make the compiler try to generate more branch-free code if possible.
If not specified the value is selected depending on the processor type that
is being compiled for.
<br><dt><code>-mzdcbranch</code><dt><code>-mno-zdcbranch</code><dd><a name="index-mzdcbranch-2500"></a><a name="index-mno_002dzdcbranch-2501"></a>Assume (do not assume) that zero displacement conditional branch instructions
<code>bt</code> and <code>bf</code> are fast. If <samp><span class="option">-mzdcbranch</span></samp> is specified, the
compiler prefers zero displacement branch code sequences. This is
enabled by default when generating code for SH4 and SH4A. It can be explicitly
disabled by specifying <samp><span class="option">-mno-zdcbranch</span></samp>.
<br><dt><code>-mcbranch-force-delay-slot</code><dd><a name="index-mcbranch_002dforce_002ddelay_002dslot-2502"></a>Force the usage of delay slots for conditional branches, which stuffs the delay
slot with a <code>nop</code> if a suitable instruction can't be found. By default
this option is disabled. It can be enabled to work around hardware bugs as
found in the original SH7055.
<br><dt><code>-mfused-madd</code><dt><code>-mno-fused-madd</code><dd><a name="index-mfused_002dmadd-2503"></a><a name="index-mno_002dfused_002dmadd-2504"></a>Generate code that uses (does not use) the floating-point multiply and
accumulate instructions. These instructions are generated by default
if hardware floating point is used. The machine-dependent
<samp><span class="option">-mfused-madd</span></samp> option is now mapped to the machine-independent
<samp><span class="option">-ffp-contract=fast</span></samp> option, and <samp><span class="option">-mno-fused-madd</span></samp> is
mapped to <samp><span class="option">-ffp-contract=off</span></samp>.
<br><dt><code>-mfsca</code><dt><code>-mno-fsca</code><dd><a name="index-mfsca-2505"></a><a name="index-mno_002dfsca-2506"></a>Allow or disallow the compiler to emit the <code>fsca</code> instruction for sine
and cosine approximations. The option <samp><span class="option">-mfsca</span></samp> must be used in
combination with <samp><span class="option">-funsafe-math-optimizations</span></samp>. It is enabled by default
when generating code for SH4A. Using <samp><span class="option">-mno-fsca</span></samp> disables sine and cosine
approximations even if <samp><span class="option">-funsafe-math-optimizations</span></samp> is in effect.
<br><dt><code>-mfsrra</code><dt><code>-mno-fsrra</code><dd><a name="index-mfsrra-2507"></a><a name="index-mno_002dfsrra-2508"></a>Allow or disallow the compiler to emit the <code>fsrra</code> instruction for
reciprocal square root approximations. The option <samp><span class="option">-mfsrra</span></samp> must be used
in combination with <samp><span class="option">-funsafe-math-optimizations</span></samp> and
<samp><span class="option">-ffinite-math-only</span></samp>. It is enabled by default when generating code for
SH4A. Using <samp><span class="option">-mno-fsrra</span></samp> disables reciprocal square root approximations
even if <samp><span class="option">-funsafe-math-optimizations</span></samp> and <samp><span class="option">-ffinite-math-only</span></samp> are
in effect.
<br><dt><code>-mpretend-cmove</code><dd><a name="index-mpretend_002dcmove-2509"></a>Prefer zero-displacement conditional branches for conditional move instruction
patterns. This can result in faster code on the SH4 processor.
</dl>
</body></html>