aboutsummaryrefslogtreecommitdiffstats
path: root/erts
AgeCommit message (Collapse)Author
2011-01-29Fix several bugs related to hibernate/3 and HiPEPaul Guyot
This commit fixes four related bugs: - calling hibernate/3 using a dynamic call would fail with badarg as hibernate/3 as a BIF was not implemented. hibernate/3 is generally provided as a Beam instruction, and code is translated to use this instruction when loaded. - calling hibernate/3 from HiPE would fail with badarg because this would call the aforementioned BIF which was not implemented. - calling hibernate/3 with some HiPE-native garbage in the process heap would randomly crash at the next garbage collect. This bug only happened in a complex, yet reproduceable scenarios, where native code calls beam code that calls hibernate/3, and the process has some garbage when being hibernated and the process generates garbage when awaken. - when entering HiPE, the process current_function can be set and be inaccurate. The fix is three folded: - hibernate_3 BIF now actually works instead of throwing a badarg. While hibernate_3 BIF was (usually) not called from BEAM, it is called from HiPE. hibernate behaviour is very close to the scheduler and this is why it is implemented as an instruction in BEAM. The fix consists in doing the actual hibernation (through the now exported erts_hibernate function) and setting the process flag to TRAP as well as the process status to P_WAITING. On BIF epilogue in both BEAM and HiPE, this status is tested on TRAP and if set, the scheduler is invoked. The i_hibernate instruction and translation code is now redundant and could be deleted. - hibernation now also empties the HiPE native stack, with a new function hipe_empty_nstack provided by Mikael Pettersson. - when entering HiPE through hipe_mode_switch, p->current is cleared, as suggested by Mikael Pettersson. p->current normally hold a pointer to the {M,F,A} of the current function if it exists. When hibernating, it is set to {erlang,hibernate,3}, and all stdlib hibernate tests (gen_server_SUITE:hibernate/1, proc_lib_suite:hibernate/1, etc.) actually rely on this information. Clearing p->current fixes the tests and avoids the surprise one might have when querying the process info of a process that hibernated and woke up in a native function. Non-regression tests are provided, a test for the dynamic call as well as a Makefile-handled duplication of the hibernate_SUITE into hibernate_native_SUITE for the HiPE case.
2011-01-17Merge branch 'bjorn/beam-loader/OTP-9030' into devBjörn Gustavsson
* bjorn/beam-loader/OTP-9030: (43 commits) c: Reduce memory footprint erl_posix_msg: Reduce memory footprint Introduce a few more variations of the move instructions Combine a move + jump sequence into the move_jump instruction Optimize and clean-up the exact equality/non-equality instructions Optimize addition of a small integer to a variable Introduce a special instruction for select_val with two values Introduce a few more specialized put_list instructions Eliminate the "put_list c n Dst" instructions Eliminate the specific move_sd instruction Eliminate use of GetArg1() in the badmatch and case_end instructions Eliminate use of GetArg2() in the i_element instruction Eliminate use of GetArg1() in the fast_element instruction Eliminate use of GetArg1() in the jump_on_val* instructions Eliminate use of GetArg1() in the select_val instruction beam_emu: Eliminate sloppy use of tmp_arg1 and tmp_arg2 beam_emu: Don't inline helper functions into process_main() beam_emu: Clean up calling of the error_handler module Simplify a select_val instruction that selects only one value Optimize creation of tuples ...
2011-01-17Introduce a few more variations of the move instructionsBjörn Gustavsson
Frequency counts show that move Const x(1) move Const x(2) are very common.
2011-01-17Combine a move + jump sequence into the move_jump instructionBjörn Gustavsson
That will save one word and small amount of time for each occurrence.
2011-01-17Optimize and clean-up the exact equality/non-equality instructionsBjörn Gustavsson
The is_eq_exact/3 and is_ne_exact/3 instructions are commonly used with one immediate or literal operand. Introduce three new specialized instructions: i_is_eq_exact_literal/3 i_is_ne_exact_immed/3 i_is_ne_exact_literal/3 The i_is_ne_exact_literal/3 instruction is not very frequently used, but its existence is justified because we removed in a a previous commit the special instruction for matching bignums and we now use i_is_ne_exact_literal/3 instead. For consistency, rename the existing is_eq_immed/3 instruction to is_eq_exact_immed/3. While at it, remove the optimization of an is_eq/3 instruction with an immediate operand because that optimization is already done by the compiler.
2011-01-17Optimize addition of a small integer to a variableBjörn Gustavsson
Introduce a new i_increment/4 to optimize the addition of a register and a small integer. This instruction saves two instruction words compared to the standard instructions (an i_fetch/2 instruction followed by a i_plus/3 instruction) and will also be slightly faster.
2011-01-17Introduce a special instruction for select_val with two valuesBjörn Gustavsson
The new instruction will save one word (because no size operand is needed), and is slightly faster. Handle select_tuple_arity in the same way.
2011-01-17Introduce a few more specialized put_list instructionsBjörn Gustavsson
2011-01-17Eliminate the "put_list c n Dst" instructionsBjörn Gustavsson
Since the literal (constant) pool was introduced in R12, the BEAM compiler will never generate a "put_list Const [] Dst" instruction (it will instead generate a "move [Const] Dst" instruction).
2011-01-17Eliminate the specific move_sd instructionBjörn Gustavsson
The move_sd specific instruction is no longer used since there are specific move instructions covering all possible permutations of operands. Also eliminate the move_cy instruction because it is almost never generated by the compiler.
2011-01-17Eliminate use of GetArg1() in the badmatch and case_end instructionsBjörn Gustavsson
Create separate instructions for each register type. The "badmatch x(0)" and "case_end x(0)" (which are very common) will only require a single word each, compared to two words when GetArg1() is used.
2011-01-17Eliminate use of GetArg2() in the i_element instructionBjörn Gustavsson
Use separate instructions for each register type.
2011-01-17Eliminate use of GetArg1() in the fast_element instructionBjörn Gustavsson
Use separate instructions for each register type.
2011-01-17Eliminate use of GetArg1() in the jump_on_val* instructionsBjörn Gustavsson
2011-01-17Eliminate use of GetArg1() in the select_val instructionBjörn Gustavsson
Instead of having one i_select_val_sfI instruction that uses the GetArg1() macro to fetch the controlling expression, use three separate instructions for each of the register types. That will save one word when selecting on the {x,0} register. It should also be slightly faster since a conditional branch is eliminated. Although it seems that the BEAM compiler will never generate a constant controlling expression (even with optimizations turned off), we still make sure that they will work by evaluating the select_val instruction at load time. Handle the select_tuple_arity instruction in the same way.
2011-01-17beam_emu: Eliminate sloppy use of tmp_arg1 and tmp_arg2Björn Gustavsson
The tmp_arg1 and tmp_arg2 variables are intended for transferring values from the fetch/2 instructions to instructions such as i_plus/3. In many places, however, tmp_arg1 and tmp_arg2 are used as general temporary variables within a single instruction. Improve the code generation by replacing sloppy use of tmp_arg1 and tmp_arg2 with block-local variables. In most cases, that will allow the temporary values to be kept in registers.
2011-01-17beam_emu: Don't inline helper functions into process_main()Björn Gustavsson
By default, GCC will inline calls to helper functions. Since process_main() is already huge, there is no reason to inline the helper functions (and some of them are used very seldom).
2011-01-17beam_emu: Clean up calling of the error_handler moduleBjörn Gustavsson
There were two separate functions (call_error_handler() and call_breakpoint_handler()) that were identical except for the name of the function in the error_handler module being called. Generalize call_error_handler() by adding a function name argument so that it can be used for both purposes. Also let the call_error_handler() return the new program counter instead of passing it in c_p->i. That slightly decrease the code size at the call site. There is also no need to use the Dispatch() macro to yet again decrease the reduction counter, because that has just been done by the call instruction that caused the execution of the call_error_handler or i_debug_breakpoint instruction.
2011-01-17Simplify a select_val instruction that selects only one valueBjörn Gustavsson
The compiler does not generate select_val instructions that only selects one value, but the loader may previously have created such an instruction when it splitted a select_val instruction that selected on bignums.
2011-01-17Optimize creation of tuplesBjörn Gustavsson
Combine the put_tuple/2 and all following put/1 instructions to one i_put_tuple/2 instruction. In general, that will reduce the number of instruction words by 50 percent. Measurements seem to indicate that the speed is about the same.
2011-01-17beam_makeops: Support jumping to common code from an instruction macroBjörn Gustavsson
2011-01-17beam_load: Run the packing engine before loading list argumentsBjörn Gustavsson
2011-01-17Allow packing of some more instructionsBjörn Gustavsson
2011-01-17Support packing of the 'I' type in a 64-bit emulatorBjörn Gustavsson
In many (not all) cases, the value for the 'I' type will fit into 32 bits.
2011-01-17beam_makeops: Refactor packing code to facilitate extensionsBjörn Gustavsson
We don't want the packable types listed in two places.
2011-01-17Eliminate redundant jump instructionsBjörn Gustavsson
2011-01-17Eliminate the special instructions for selecting floats and bignumsBjörn Gustavsson
2011-01-17Add test for non-matching big numberBjörn Gustavsson
2011-01-17BEAM loader: Introduce a new move2_xxxx instructionBjörn Gustavsson
2011-01-17BEAM loader: Pack more instructions using a new 'Q' typeBjörn Gustavsson
Introduce a new 'Q' type, similar to 'P' except that it can be packed.
2011-01-17If the wordsize is 64 bits, pack up to 4 operands into a wordBjörn Gustavsson
In the 32-bit BEAM emulator, it is only possible to pack 3 register operands into one word. Therefore, the move2 instruction (that has 4 operands) needs two words for its operands. Take advantage of the larger wordsize in the 64-bit emulator and pack up to 4 operands into a single word.
2011-01-17Pass the external word size to the beam_makeops scriptBjörn Gustavsson
Giving the beam_makeops script access to the external word size (=the size of instruction words) will allow it to pack more operands into a word for the 64 bits emulator.
2011-01-17BEAM loader: Combine is_type/1 and is_eq/1 instructionsBjörn Gustavsson
In the transformation engine in the loader, an is_eq/1 instruction is currently always preceded by an is_type/1 instruction. Therefore, save a word and slight amount of time by combining those instructions into an is_type_eq/2 instruction.
2011-01-17BEAM loader: Omit type tests for instructions that don't need themBjörn Gustavsson
2011-01-17beam_debug: Fix dissambly of some variable-operand instructionsBjörn Gustavsson
The i_jump_on_val_zero/3 and i_select_tuple_arity/3 instructions were not disassembled correctly.
2011-01-17beam_debug: Change one occurrence of "X[0]" to "x[0]" for consistencyBjörn Gustavsson
2011-01-17beam_makeops: Allow -pack and -nonext to be combinedBjörn Gustavsson
2011-01-17beam_makeops: Relax requirements for having specific instructionsBjörn Gustavsson
2011-01-17BEAM loader: Fix bug in handling of "rest" argumentsBjörn Gustavsson
It would only really work in simple case like: select_val S=q Fail=f Size=u Rest=* => ... where all operands for a single instruction where bound to variables, and not for more complicated cases such as: i_put_tuple Dst Arity Puts=* | put PutSrc => ...
2011-01-17Remove the last vestiges of the allocating fmove/2 instructionBjörn Gustavsson
There was a version of the BEAM loader and emulator that had two versions of the fmove/2 instruction, one version that allocated heap space internally and a newer version that assumed that a previous test_heap/2 instruction had already allocated the heap space. Though the allocating fmove/2 instruction is no longer supported, some vestiges of it still remains.
2011-01-17Makefile.in: Add dependency for files generated by util/beam_makeopsBjörn Gustavsson
2011-01-17Add utils/count for counting the static instruction frequencyBjörn Gustavsson
2011-01-17Add erts_debug:instructions/0 for listing all specific instructionsBjörn Gustavsson
erts_debug:instructions/0 is useful for finding which specific instructions that are not used at all.
2011-01-13Merge branch 'pan/fortify-source-gcc-4.5' into devPatrik Nyblom
* pan/fortify-source-gcc-4.5: Remove strange name field in efile_drv.c as it upsets -D_FORTIFY_SOURCE OTP-9025
2011-01-12Remove strange name field in efile_drv.c as it upsets -D_FORTIFY_SOURCEPatrik Nyblom
2011-01-11Remove special treatment of EADDRINUSEPatrik Nyblom
2011-01-11check return value of epmd server listen() callSteve Vinoski
The listen() call can fail due to a variety of conditions, so check its return value and if it fails, print a suitable debug message if appropriate and then exit. The exit values used are the same for those already used for bind() failures: 0 if the error is EADDRINUSE, 1 otherwise.
2010-12-29Merge branch 'egil/timer-wheel-min-time/OTP-8990' into devBjörn-Egil Dahlberg
* egil/timer-wheel-min-time/OTP-8990: Teach timer-wheel slots to use double linked lists Remove timer-thread implementation Refactor timer interface Teach timer-wheel to keep min time
2010-12-29Teach timer-wheel slots to use double linked listsBjörn-Egil Dahlberg
Conflicts: erts/emulator/beam/erl_time.h
2010-12-21Merge branch 'bjorn/bs-zero-width-bug/OTP-8997' into devBjörn Gustavsson
* bjorn/bs-zero-width-bug/OTP-8997: Fix type-checking of variable used in zero-width bit syntax construction