arno/hashcat
1
0
Fork 0
mirror of https://github.com/hashcat/hashcat.git synced 2025-07-07 15:18:15 +00:00
Code Issues Releases Wiki Activity

Metal Backend:

Browse Source 
- added support to 2D/3D Compute
- improved compute workloads calculation
Makefile:
- updated MACOSX_DEPLOYMENT_TARGET to 15.0
Unit tests:
- updated install_modules.sh with Crypt::Argon2

Argon2 start works with Apple Metal
This commit is contained in:
Gabriele Gristina 2025-07-03 22:06:32 +02:00
parent b9b20c3340
commit bcc351068f
No known key found for this signature in database
GPG Key ID: 9F68B59298F311F0
11 changed files with 93 additions and 41 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
OpenCL/inc_amp.h
View File

@ -16,7 +16,7 @@
GLOBAL_AS const bf_t *bfs_buf, \
CONSTANT_AS const u32 &combs_mode, \
CONSTANT_AS const u64 &gid_max, \
uint hc_gid [[ thread_position_in_grid ]]
uint3 hc_gid [[ thread_position_in_grid ]]
#else // CUDA, HIP, OpenCL

8
OpenCL/inc_common.h
View File

@ -124,10 +124,10 @@
#if defined IS_METAL
#define KERN_ATTR_MAIN_PARAMS \
uint hc_gid [[ thread_position_in_grid ]], \
uint hc_lid [[ thread_position_in_threadgroup ]], \
uint hc_lsz [[ threads_per_threadgroup ]], \
uint hc_bid [[ threadgroup_position_in_grid ]]
uint3 hc_gid [[ thread_position_in_grid ]], \
uint3 hc_lid [[ thread_position_in_threadgroup ]], \
uint3 hc_lsz [[ threads_per_threadgroup ]], \
uint3 hc_bid [[ threadgroup_position_in_grid ]]
#endif // IS_METAL
/*

6
OpenCL/inc_markov.h
View File

@ -19,7 +19,7 @@
CONSTANT_AS const u32 &bits14, \
CONSTANT_AS const u32 &bits15, \
CONSTANT_AS const u64 &gid_max, \
uint hc_gid [[ thread_position_in_grid ]]
uint3 hc_gid [[ thread_position_in_grid ]]
#define KERN_ATTR_R_MARKOV \
GLOBAL_AS bf_t *pws_buf_r, \
@ -31,7 +31,7 @@
CONSTANT_AS const u32 &bits14, \
CONSTANT_AS const u32 &bits15, \
CONSTANT_AS const u64 &gid_max, \
uint hc_gid [[ thread_position_in_grid ]]
uint3 hc_gid [[ thread_position_in_grid ]]
#define KERN_ATTR_C_MARKOV \
GLOBAL_AS pw_t *pws_buf, \
@ -43,7 +43,7 @@
CONSTANT_AS const u32 &bits14, \
CONSTANT_AS const u32 &bits15, \
CONSTANT_AS const u64 &gid_max, \
uint hc_gid [[ thread_position_in_grid ]]
uint3 hc_gid [[ thread_position_in_grid ]]
#else // CUDA, HIP, OpenCL

23
OpenCL/inc_platform.h
View File

@ -73,10 +73,25 @@ DECLSPEC u32 hc_atomic_dec (volatile GLOBAL_AS u32 *p);
DECLSPEC u32 hc_atomic_inc (volatile GLOBAL_AS u32 *p);
DECLSPEC u32 hc_atomic_or (volatile GLOBAL_AS u32 *p, volatile const u32 val);
#define get_global_id(param) hc_gid
#define get_local_id(param) hc_lid
#define get_local_size(param) hc_lsz
#define get_group_id(param) hc_bid
#define get_global_id(dimindx) \
((dimindx) == 0 ? hc_gid.x : \
(dimindx) == 1 ? hc_gid.y : \
(dimindx) == 2 ? hc_gid.z : -1)
#define get_group_id(dimindx) \
((dimindx) == 0 ? hc_bid.x : \
(dimindx) == 1 ? hc_bid.y : \
(dimindx) == 2 ? hc_bid.z : -1)
#define get_local_id(dimindx) \
((dimindx) == 0 ? hc_lid.x : \
(dimindx) == 1 ? hc_lid.y : \
(dimindx) == 2 ? hc_lid.z : -1)
#define get_local_size(dimindx) \
((dimindx) == 0 ? hc_lsz.x : \
(dimindx) == 1 ? hc_lsz.y : \
(dimindx) == 2 ? hc_lsz.z : -1)
DECLSPEC u32x rotl32 (const u32x a, const int n);
DECLSPEC u32x rotr32 (const u32x a, const int n);

10
OpenCL/inc_shared.h
View File

@ -13,28 +13,28 @@
GLOBAL_AS u32 *pws_comp, \
GLOBAL_AS pw_t *pws_buf, \
CONSTANT_AS const u64 &gid_max, \
uint hc_gid [[ thread_position_in_grid ]]
uint3 hc_gid [[ thread_position_in_grid ]]
#define KERN_ATTR_GPU_MEMSET \
GLOBAL_AS uint4 *buf, \
CONSTANT_AS const u32 &value, \
CONSTANT_AS const u64 &gid_max, \
uint hc_gid [[ thread_position_in_grid ]]
uint3 hc_gid [[ thread_position_in_grid ]]
#define KERN_ATTR_GPU_BZERO \
GLOBAL_AS uint4 *buf, \
CONSTANT_AS const u64 &gid_max, \
uint hc_gid [[ thread_position_in_grid ]]
uint3 hc_gid [[ thread_position_in_grid ]]
#define KERN_ATTR_GPU_ATINIT \
GLOBAL_AS pw_t *buf, \
CONSTANT_AS const u64 &gid_max, \
uint hc_gid [[ thread_position_in_grid ]]
uint3 hc_gid [[ thread_position_in_grid ]]
#define KERN_ATTR_GPU_UTF8_TO_UTF16 \
GLOBAL_AS pw_t *pws_buf, \
CONSTANT_AS const u64 &gid_max, \
uint hc_gid [[ thread_position_in_grid ]]
uint3 hc_gid [[ thread_position_in_grid ]]
#else // CUDA, HIP, OpenCL

4
docs/changes.txt
View File

@ -148,19 +148,23 @@
- Status Code: Add specific return code for self-test fail (-11)
- Scrypt: Increase buffer sizes in module for hash mode 8900 to allow longer scrypt digests
- Unicode: Update UTF-8 to UTF-16 conversion to match RFC 3629
- Unit tests: Updated install_modules.sh with Crypt::Argon2
- User Options: Added error message when mixing --username and --show to warn users of exponential delay
- MetaMask: update extraction tool to support MetaMask Mobile wallets
- SecureCRT MasterPassphrase v2: update module, pure kernels and test unit. Add optimized kernels.
- Metal Backend: added workaround to prevent 'Infinite Loop' bug when build kernels
- Metal Backend: added workaround to set the true Processor value in Metal devices on Apple Intel
- Metal Backend: added support to 2D/3D Compute
- Metal Backend: allow use of devices with Metal if runtime version is >= 200
- Metal Backend: disable Metal devices only if at least one OpenCL device is active
- Metal Backend: improved compute workloads calculation
- Modules: Check UnpackSize to raise false positive with hc_decompress_rar
- User Options: added --metal-compiler-runtime option
- Hardware Monitor: avoid sprintf in src/ext_iokit.c
- Hardware Monitor: Splitting hwmon_ctx_init function into smaller library-specific functions
- Help: show supported hash-modes only with -hh
- Makefile: prevent make failure with Apple Silicon in case of partial rebuild
- Makefile: updated MACOSX_DEPLOYMENT_TARGET to 15.0
- Rules: Rename best64.rule to best66.rule and remove the unknown section from it
* changes v6.2.5 -> v6.2.6

2
include/ext_metal.h
View File

@ -111,7 +111,7 @@ int hc_mtlCreateLibraryWithFile (void *hashcat_ctx, mtl_device_id metal_devi
int hc_mtlEncodeComputeCommand_pre (void *hashcat_ctx, mtl_pipeline metal_pipeline, mtl_command_queue metal_command_queue, mtl_command_buffer *metal_command_buffer, mtl_command_encoder *metal_command_encoder);
int hc_mtlSetCommandEncoderArg (void *hashcat_ctx, mtl_command_encoder metal_command_encoder, size_t off, size_t idx, mtl_mem buf, void *host_data, size_t host_data_size);
int hc_mtlEncodeComputeCommand (void *hashcat_ctx, mtl_command_encoder metal_command_encoder, mtl_command_buffer metal_command_buffer, const size_t global_work_size[3], const size_t local_work_size[3], double *ms);
int hc_mtlEncodeComputeCommand (void *hashcat_ctx, mtl_command_encoder metal_command_encoder, mtl_command_buffer metal_command_buffer, const unsigned int work_dim, const size_t global_work_size[3], const size_t local_work_size[3], double *ms);
#endif // __APPLE__

2
src/Makefile
View File

@ -358,7 +358,7 @@ LFLAGS_NATIVE += -lpthread
endif # NetBSD
ifeq ($(UNAME),Darwin)
export MACOSX_DEPLOYMENT_TARGET=10.15
export MACOSX_DEPLOYMENT_TARGET=15.0
CFLAGS_NATIVE := $(CFLAGS)
CFLAGS_NATIVE += -DWITH_HWMON

23
src/backend.c
View File

@ -2206,7 +2206,7 @@ int run_metal_kernel_atinit (hashcat_ctx_t *hashcat_ctx, hc_device_param_t *devi
double ms = 0;
if (hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, global_work_size, local_work_size, &ms) == -1) return -1;
if (hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, 1, global_work_size, local_work_size, &ms) == -1) return -1;
return 0;
}
@ -2234,7 +2234,7 @@ int run_metal_kernel_utf8toutf16le (hashcat_ctx_t *hashcat_ctx, hc_device_param_
double ms = 0;
if (hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, global_work_size, local_work_size, &ms) == -1) return -1;
if (hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, 1, global_work_size, local_work_size, &ms) == -1) return -1;
return 0;
}
@ -2265,7 +2265,7 @@ int run_metal_kernel_bzero (hashcat_ctx_t *hashcat_ctx, hc_device_param_t *devic
double ms = 0;
if (hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, global_work_size, local_work_size, &ms) == -1) return -1;
if (hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, 1, global_work_size, local_work_size, &ms) == -1) return -1;
}
if (num16m)
@ -2910,29 +2910,34 @@ int run_kernel (hashcat_ctx_t *hashcat_ctx, hc_device_param_t *device_param, con
num_elements = num_elements * kernel_threads;
}
unsigned int work_dim = 1;
size_t global_work_size[3] = { num_elements, 1, 1 };
size_t local_work_size[3] = { kernel_threads, 1, 1 };
if ((hashconfig->opti_type & OPTI_TYPE_SLOW_HASH_DIMY_INIT) && (kern_run == KERN_RUN_1))
{
global_work_size[1] = local_work_size[1] = hashcat_ctx->hashes->salts_buf->salt_dimy;
work_dim = 2;
}
if ((hashconfig->opti_type & OPTI_TYPE_SLOW_HASH_DIMY_LOOP) && (kern_run == KERN_RUN_2))
{
global_work_size[1] = local_work_size[1] = hashcat_ctx->hashes->salts_buf->salt_dimy;
work_dim = 2;
}
if ((hashconfig->opti_type & OPTI_TYPE_SLOW_HASH_DIMY_COMP) && (kern_run == KERN_RUN_3))
{
global_work_size[1] = local_work_size[1] = hashcat_ctx->hashes->salts_buf->salt_dimy;
work_dim = 2;
}
double ms = 0;
if (is_autotune == true)
{
hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, global_work_size, local_work_size, &ms);
hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, work_dim, global_work_size, local_work_size, &ms);
// hc_mtlEncodeComputeCommand_pre() must be called before every hc_mtlEncodeComputeCommand()
if (hc_mtlEncodeComputeCommand_pre (hashcat_ctx, metal_pipeline, device_param->metal_command_queue, &metal_command_buffer, &metal_command_encoder) == -1) return -1;
@ -2951,7 +2956,7 @@ int run_kernel (hashcat_ctx_t *hashcat_ctx, hc_device_param_t *device_param, con
}
}
const int rc_cc = hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, global_work_size, local_work_size, &ms);
const int rc_cc = hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, work_dim, global_work_size, local_work_size, &ms);
if (rc_cc != -1)
{
@ -3344,7 +3349,7 @@ int run_kernel_mp (hashcat_ctx_t *hashcat_ctx, hc_device_param_t *device_param,
double ms = 0;
if (hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, global_work_size, local_work_size, &ms) == -1) return -1;
if (hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, 1, global_work_size, local_work_size, &ms) == -1) return -1;
}
#endif // __APPLE__
@ -3435,7 +3440,7 @@ int run_kernel_tm (hashcat_ctx_t *hashcat_ctx, hc_device_param_t *device_param)
double ms = 0;
if (hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, global_work_size, local_work_size, &ms) == -1) return -1;
if (hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, 1, global_work_size, local_work_size, &ms) == -1) return -1;
}
#endif // __APPLE__
@ -3519,7 +3524,7 @@ int run_kernel_amp (hashcat_ctx_t *hashcat_ctx, hc_device_param_t *device_param,
double ms = 0;
const int rc_cc = hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, global_work_size, local_work_size, &ms);
const int rc_cc = hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, 1, global_work_size, local_work_size, &ms);
// release tmp_buf
@ -3599,7 +3604,7 @@ int run_kernel_decompress (hashcat_ctx_t *hashcat_ctx, hc_device_param_t *device
double ms = 0;
if (hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, global_work_size, local_work_size, &ms) == -1) return -1;
if (hc_mtlEncodeComputeCommand (hashcat_ctx, metal_command_encoder, metal_command_buffer, 1, global_work_size, local_work_size, &ms) == -1) return -1;
}
#endif // __APPLE__

39
src/ext_metal.m
View File

@ -195,11 +195,14 @@ static int hc_mtlBuildOptionsToDict (void *hashcat_ctx, const char *build_option
}
// if set, add INCLUDE_PATH to hack Apple kernel build from source limitation on -I usage
if (include_path != nil)
{
NSString *path_key = @"INCLUDE_PATH";
NSString *path_value = [NSString stringWithCString: include_path encoding: NSUTF8StringEncoding];
// Include path may contain spaces, escape them with a backslash
path_value = [path_value stringByReplacingOccurrencesOfString:@" " withString:@"\\ "];
[build_options_dict setObject:path_value forKey:path_key];
@ -743,6 +746,7 @@ int hc_mtlCreateKernel (void *hashcat_ctx, mtl_device_id metal_device, mtl_libra
dispatch_queue_t queue = dispatch_get_global_queue (DISPATCH_QUEUE_PRIORITY_DEFAULT, 0);
// if no user-defined runtime, set to METAL_COMPILER_RUNTIME
long timeout = (user_options->metal_compiler_runtime > 0) ? user_options->metal_compiler_runtime : METAL_COMPILER_RUNTIME;
dispatch_time_t when = dispatch_time (DISPATCH_TIME_NOW,NSEC_PER_SEC * timeout);
@ -1314,10 +1318,21 @@ int hc_mtlSetCommandEncoderArg (void *hashcat_ctx, mtl_command_encoder metal_com
return 0;
}
int hc_mtlEncodeComputeCommand (void *hashcat_ctx, mtl_command_encoder metal_command_encoder, mtl_command_buffer metal_command_buffer, const size_t global_work_size[3], const size_t local_work_size[3], double *ms)
int hc_mtlEncodeComputeCommand (void *hashcat_ctx, mtl_command_encoder metal_command_encoder, mtl_command_buffer metal_command_buffer, const unsigned int work_dim, const size_t global_work_size[3], const size_t local_work_size[3], double *ms)
{
MTLSize numThreadgroups = {local_work_size[0], local_work_size[1], local_work_size[2]};
MTLSize threadsGroup = {global_work_size[0], global_work_size[1], global_work_size[2]};
MTLSize threadsPerThreadgroup =
{
local_work_size[0],
local_work_size[1],
local_work_size[2]
};
MTLSize threadgroupsPerGrid =
{
(global_work_size[0] + threadsPerThreadgroup.width - 1) / threadsPerThreadgroup.width,
work_dim > 1 ? (global_work_size[1] + threadsPerThreadgroup.height - 1) / threadsPerThreadgroup.height : 1,
work_dim > 2 ? (global_work_size[2] + threadsPerThreadgroup.depth - 1) / threadsPerThreadgroup.depth : 1
};
if (metal_command_encoder == nil)
{
@ -1333,7 +1348,7 @@ int hc_mtlEncodeComputeCommand (void *hashcat_ctx, mtl_command_encoder metal_com
return -1;
}
[metal_command_encoder dispatchThreadgroups: threadsGroup threadsPerThreadgroup: numThreadgroups];
[metal_command_encoder dispatchThreadgroups: threadgroupsPerGrid threadsPerThreadgroup: threadsPerThreadgroup];
[metal_command_encoder endEncoding];
[metal_command_buffer commit];
@ -1377,7 +1392,11 @@ int hc_mtlCreateLibraryWithFile (void *hashcat_ctx, mtl_device_id metal_device,
if (k_string != nil)
{
id <MTLLibrary> r = [metal_device newLibraryWithFile: k_string error: &error];
NSURL *libURL = [NSURL fileURLWithPath: k_string];
if (libURL != nil)
{
id <MTLLibrary> r = [metal_device newLibraryWithURL: libURL error:&error];
if (error != nil)
{
@ -1389,6 +1408,7 @@ int hc_mtlCreateLibraryWithFile (void *hashcat_ctx, mtl_device_id metal_device,
return 0;
}
}
return -1;
}
@ -1420,10 +1440,17 @@ int hc_mtlCreateLibraryWithSource (void *hashcat_ctx, mtl_device_id metal_device
}
compileOptions.preprocessorMacros = build_options_dict;
/*
compileOptions.optimizationLevel = MTLLibraryOptimizationLevelSize;
compileOptions.mathMode = MTLMathModeSafe;
// compileOptions.mathMode = MTLMathModeRelaxed;
// compileOptions.enableLogging = true;
// compileOptions.fastMathEnabled = false;
*/
}
// todo: detect current os version and choose the right
// compileOptions.languageVersion = MTL_LANGUAGEVERSION_2_3;
// compileOptions.languageVersion = MTL_LANGUAGEVERSION_2_3;
/*
if (@available(macOS 12.0, *))
{

1
tools/install_modules.sh
View File

@ -18,6 +18,7 @@ cpan install Authen::Passphrase::LANManager \
Bitcoin::Crypto::Base58 \
Compress::Zlib \
Convert::EBCDIC \
Crypt::Argon2 \
Crypt::AuthEnc::GCM \
Crypt::Camellia \
Crypt::CBC \