Merge remote-tracking branch 'origin/GP-49_external_disassembly_upgrade--SQUASHED' into Ghidra_9.2

GPL/GnuDisassembler/src/gdis/c/disasm_1.c

@@ -21,7 +21,7 @@ void listSupportedArchMachTargets(void)
 	const char** targetList;
 	const char** archList;
 	int i, j;
-	
+
 	targetList = bfd_target_list();
 	if(targetList != NULL){
 		for(i=0, j=0; targetList[i] !=0; i++){
@@ -29,7 +29,7 @@ void listSupportedArchMachTargets(void)
 		}
 	}
 	printf("\ndone with targetList.\n");
-	
+
 	archList = bfd_arch_list();
 	if(archList != NULL){
 		for(i=0, j=0; archList[i] !=0; i++){
@@ -44,29 +44,29 @@ void listSupportedArchMachTargets(void)
 /* sprintf to a "stream".  */
 int objdump_sprintf (SFILE *f, const char *format, ...)
 {
-	
+
 	int i;
 	size_t n;
 	va_list args;
- 
+
 	va_start (args, format);
 	n = vsnprintf (f->buffer + f->pos, BUFF_SIZE, format, args);
 	strncat(disassembled_buffer, f->buffer, n);
 	va_end (args);
-  
+
 	return n;
 }
 
 
 void configureDisassembleInfo(bfd* abfd,
-							  disassemble_info* info, 
+		disassemble_info* info,
-							  enum bfd_architecture arch, 
+		enum bfd_architecture arch,
-							  unsigned long mach, 
+		unsigned long mach,
-							  enum bfd_endian end)
+		enum bfd_endian end)
 {
 
 	memset(sfile.buffer, 0x00, BUFF_SIZE);
-	
+
 	INIT_DISASSEMBLE_INFO(*info, stdout, objdump_sprintf);
 	info->arch = (enum bfd_architecture) arch;
 	info->mach = mach;
@@ -79,24 +79,24 @@ void configureDisassembleInfo(bfd* abfd,
 }
 
 disassembler_ftype configureBfd(bfd* abfd, 
-								enum bfd_architecture arch, 
+		enum bfd_architecture arch,
-								unsigned long mach, 
+		unsigned long mach,
-								enum bfd_endian endian, 
+		enum bfd_endian endian,
-								disassemble_info* DI,
+		disassemble_info* DI,
-								disassembler_ftype* disassemble_fn)
+		disassembler_ftype* disassemble_fn)
 {	
 	struct bfd_target *xvec;
-	
+
 	abfd->flags |= EXEC_P;
-	
+
-	
+
 	// set up xvec byteorder. 
 	xvec = (struct bfd_target *) malloc (sizeof (struct bfd_target));
 	memset(xvec, 0x00, sizeof (struct bfd_target));	
 	memcpy (xvec, abfd->xvec, sizeof (struct bfd_target));
 	xvec->byteorder = endian;
 	abfd->xvec = xvec;
-	
+
 	configureDisassembleInfo(abfd, DI, arch, mach, endian);
 	if(endian == BFD_ENDIAN_BIG){
 		bfd_big_endian(abfd);
@@ -106,29 +106,29 @@ disassembler_ftype configureBfd(bfd* abfd,
 		bfd_little_endian(abfd);
 		DI->display_endian = DI->endian = BFD_ENDIAN_LITTLE;
 	}
-	
+
 	/*
 	bfd_error_type err = bfd_get_error();
 	printf("bfd_error_msg: %s.\n", bfd_errmsg(err));
-	*/
+	 */
-	
+
 	/* Use libopcodes to locate a suitable disassembler.  */
 	*disassemble_fn = NULL;
 	*disassemble_fn = disassembler (arch, endian == BFD_ENDIAN_BIG, mach, abfd);
 	if (!*disassemble_fn){
 		printf("can't disassemble for arch 0x%08X, mach 0x%08lX\n", arch, mach);
 		exit(1);
-    }
+	}
-    
+
-    return *disassemble_fn;
+	return *disassemble_fn;
 }
 
 
 
 int disassemble_buffer( disassembler_ftype disassemble_fn,
-						disassemble_info *info,
+		disassemble_info *info,
-						int* offset,						
+		int* offset,
-						PDIS_INFO pDisInfo)
+		PDIS_INFO pDisInfo)
 {
 	int i, j, size = 0;
 	int len = 0;
@@ -136,7 +136,7 @@ int disassemble_buffer( disassembler_ftype disassemble_fn,
 	while ( *offset < info->buffer_length ) {
 		/* call the libopcodes disassembler */
 		memset(pDisInfo->disassemblyString, 0x00, MAX_DIS_STRING);
-		
+
 		/* set the insn_info_valid bit to 0, as explained in BFD's
 		 * include/dis-asm.h. The bit will then be set to tell us
 		 * whether the decoder supports "extra" information about the
@@ -147,7 +147,7 @@ int disassemble_buffer( disassembler_ftype disassemble_fn,
 		size = (*disassemble_fn)(info->buffer_vma + *offset, info);
 		/* -- analyze disassembled instruction here -- */
 		/* -- print any symbol names as labels here -- */
-		
+
 		/* save off corresponding hex bytes */
 		for ( j= 0,i = 0; i < 8; i++, j+=3) {
 			if ( i < size ){				
@@ -155,7 +155,7 @@ int disassemble_buffer( disassembler_ftype disassemble_fn,
 				pDisInfo->bytesBufferBin[i] = info->buffer[*offset + i];
 			}
 		}
-		
+
 		/* add the augmented information to our disassembly info struct */
 		pDisInfo->count = size;
 		pDisInfo->insn_info_valid = info->insn_info_valid;
@@ -167,44 +167,44 @@ int disassemble_buffer( disassembler_ftype disassemble_fn,
 
 		strcat(&(pDisInfo->disassemblyString[0]), disassembled_buffer);
 		memset(disassembled_buffer, 0x00, BUFF_SIZE);
-		
+
 		if(size != 0){
 			*offset += size; /* advance position in buffer */
 			goto END;
 		}
 	}
 
-END:	
+	END:
 	return size;
 }
 
 void processBuffer(unsigned char* buff, 
-		   int buff_len,
+		int buff_len,
-		   bfd_vma buff_vma,
+		bfd_vma buff_vma,
-		   disassembler_ftype disassemble_fn, 
+		disassembler_ftype disassemble_fn,
-		   struct disassemble_info* DI)
+		struct disassemble_info* DI)
 {
 	int bytesConsumed = -1;
 	int offset = 0;
 	int numDisassemblies = 0;
 	int i;
-		
+
 	DI->buffer = buff; /* buffer of bytes to disassemble */
 	DI->buffer_length = buff_len; /* size of buffer */
 	DI->buffer_vma = buff_vma; /* base RVA of buffer */
-								 
+
 	memset(disassemblyInfoBuffer, 0x00, sizeof(DIS_INFO)*MAX_NUM_ENTRIES);
-			
+
 	while((buff_len - offset) > 0 && bytesConsumed != 0 && numDisassemblies < MAX_NUM_ENTRIES){
 		bytesConsumed = disassemble_buffer( disassemble_fn, DI, &offset, &(disassemblyInfoBuffer[numDisassemblies++]));	
 	}
 	for (i = 0; i < numDisassemblies; i++) {
-	    printf("%s\nInfo: %d,%d,%d,%d,%d\n", disassemblyInfoBuffer[i].disassemblyString,
+		printf("%s\nInfo: %d,%d,%d,%d,%d\n", disassemblyInfoBuffer[i].disassemblyString,
-		   disassemblyInfoBuffer[i].count, 
+				disassemblyInfoBuffer[i].count,
-		   disassemblyInfoBuffer[i].insn_info_valid, 
+				disassemblyInfoBuffer[i].insn_info_valid,
-		   disassemblyInfoBuffer[i].branch_delay_insns,
+				disassemblyInfoBuffer[i].branch_delay_insns,
-		   disassemblyInfoBuffer[i].data_size,
+				disassemblyInfoBuffer[i].data_size,
-		   disassemblyInfoBuffer[i].insn_type);
+				disassemblyInfoBuffer[i].insn_type);
 	}
 }
 
@@ -224,7 +224,7 @@ int main(int argc, char* argv[]){
 	char elf_file_location[MAX_ELF_FILE_PATH_LEN];
 	char arch_str[256];
 	char mach_str[256];
-		
+
 	if ( argc < 8) {
 		fprintf(stderr, "Usage: %s target-str, arch, mach, disassembly base-addr (for rel offsets instrs), full-path to Little and Big Elfs, big/little ascii-byte-string up to %d chars\n", argv[0], MAX_ASCII_CHAR_BYTE_STRING);
 		listSupportedArchMachTargets();
@@ -247,16 +247,16 @@ int main(int argc, char* argv[]){
 	mach = 0x00000000;
 	arch = (enum bfd_architecture) 0x00;
 	offset = 0x00000000;
-	
+
 	sscanf(argv[2], "%128s", arch_str);
 	sscanf(argv[3], "%18lX", &mach);
 	sscanf(argv[4], "%10X", &end);
 	sscanf(argv[5], "%18lX", &offset);
-	
+
 	// if arch starts with 0x, then parse a number
 	//  else lookup the string in the table to get the arch, ignore the mach
 	if (arch_str[0] == '0' && arch_str[1] == 'x') {
-	   sscanf(arch_str, "%10X", &arch);
+		sscanf(arch_str, "%10X", &arch);
 	} else {
 		const char** archList = bfd_arch_list();
 		const bfd_arch_info_type* ait;
@@ -271,17 +271,17 @@ int main(int argc, char* argv[]){
 			archList++;
 		}
 		if (ait == NULL) {
-		    printf("Couldn't find arch %s\n", arch_str);
+			printf("Couldn't find arch %s\n", arch_str);
-		    return(-1);
+			return(-1);
 		}
 	}
-	    
+
-	
+
 	endian = (enum bfd_endian) end;
 	/* open a correct type of file to fill in most of the required data. */
-	
+
 	//	printf("Arch is: 0x%08X, Machine is: 0x%08lX Endian is: 0x%02X.\n", arch, mach, endian);
-	
+
 	memset(elf_file_location, 0x00, MAX_ELF_FILE_PATH_LEN);
 	strncpy(elf_file_location, argv[6], MAX_ELF_FILE_PATH_LEN-sizeof(LITTLE_ELF_FILE)-2); // actual file name and nulls
 
@@ -294,59 +294,66 @@ int main(int argc, char* argv[]){
 		stdin_mode = 1; // use STDIN
 	}
 
-	unsigned char byteBuffer[BYTE_BUFFER_SIZE];
-	char byteStringBuffer[(BYTE_BUFFER_SIZE*2)];
-
-	char addressStringBuffer[128];
-
 	if (endian == BFD_ENDIAN_BIG){
 		strcat(elf_file_location, BIG_ELF_FILE);
 	}
 	else {
 		strcat(elf_file_location, LITTLE_ELF_FILE);
 	}
-	
+
+	bfd_init();
+
 	while (stdin_mode) {
-	
+
+		//bfd_init();
 		// convert user input AsciiHex to Binary data for processing
 		char tmp[3];
 		unsigned int byteValue;
 		tmp[0] = tmp[1]  = tmp[2] = 0x00;
+		char disassemblerOptions[128] = {0};
+		unsigned char byteBuffer[BYTE_BUFFER_SIZE] = {0};
+		char byteStringBuffer[(BYTE_BUFFER_SIZE*2)] = {0};
+		char addressStringBuffer[128] = {0};
+		char bytesAndOptionsBuffer[BYTE_BUFFER_SIZE*2 + sizeof(disassemblerOptions)] = {0};
 
 		if (stdin_mode == 1) { // use stdin
 			// read in the address
 			if (fgets(addressStringBuffer, sizeof(addressStringBuffer), stdin)) {
-
-			    //fprintf(stderr, "read: %s\n", addressStringBuffer);
-				//char *p = strchr(addressStringBuffer, '\n');
-				//if (p) {
-				//	*p = '\0';
-				//}
-
 				sscanf(addressStringBuffer, "%18lX", &offset);
 			}
-			//getchar();
+			//Read in the rest of the input. There are two possible styles:
-			// read in the ASCII hex string from stdin
+			//"old": input is just the bytes to disassemble
-			if (fgets(byteStringBuffer, sizeof(byteStringBuffer), stdin)) {
+		    //"new": input = bytes"*"<disassmbler_options>
-
+			//for compatibility reasons, handle both styles
-			    //fprintf(stderr, "read: %s\n", byteStringBuffer);
+			if (fgets(bytesAndOptionsBuffer, sizeof(bytesAndOptionsBuffer), stdin)) {
-				// remove trailing newline 
+				char *p = strchr(bytesAndOptionsBuffer, '*');
-				char *p = strchr(byteStringBuffer, '\n');
 				if (p) {
-					*p = '\0';
+					strncpy(byteStringBuffer,bytesAndOptionsBuffer,(int) (p - bytesAndOptionsBuffer));
+					strncpy(disassemblerOptions,p+1,sizeof(disassemblerOptions));
+					p = strchr(disassemblerOptions,'\n');
+					if (p){
+						*p = '\0';
+					}
+				}
+				else {
+					//no "*" in the string, so no disassembly options
+					//replace newline with null terminator
+					strncpy(byteStringBuffer, bytesAndOptionsBuffer, sizeof(byteStringBuffer));
+					p = strchr(byteStringBuffer,'\n');
+					if (p){
+						*p = '\0';
+					}
 				}
-				//if (strcmp(byteStringBuffer, "EOF") == 0) {
-				//return 0; // terminate on EOF string
-				//}
-			} else {
-			    fprintf(stderr, "exiting, no ASCII hex found\n");
-			    return 0; // finished! #TODO
 			}
-
+			else {
-		} else {
+				fprintf(stderr, "exiting, no ASCII hex found\n");
+				return 0; // finished! #TODO
+			}
+		}
+		else {
 			if(strlen(byteString) > BYTE_BUFFER_SIZE*2) {
-				fprintf(stderr, "Max ascii string size is %d you provided: %lu chars. Exiting.\n", BYTE_BUFFER_SIZE*2, 
+				fprintf(stderr, "Max ascii string size is %d you provided: %lu chars. Exiting.\n", BYTE_BUFFER_SIZE*2,
-				       strlen(byteString));
+						strlen(byteString));
 				exit(-1);
 			}
 			strncpy(byteStringBuffer, byteString, BYTE_BUFFER_SIZE*2);
@@ -378,9 +385,9 @@ int main(int argc, char* argv[]){
 		  for(j=0; j < BYTE_BUFFER_SIZE; j++){
 		  printf("0x%02X ", byteBuffer[j]);
 		  }
-		*/
+		 */
-	
+
-		bfd_init( );
+		//bfd_init( );
 		target = argv[1];
 		bfd_set_default_target(target);
 
@@ -392,7 +399,7 @@ int main(int argc, char* argv[]){
 			bfdfile = bfd_openr(elf_file_location, target );
 			if ( ! bfdfile ) {
 				printf("Error opening BIG ELF file: %s\n", elf_file_location);
-				bfd_perror( "Error on bfdfile" );	
+				bfd_perror( "Error on bfdfile" );
 				return(3);
 			}
 		}
@@ -401,34 +408,37 @@ int main(int argc, char* argv[]){
 			if ( ! bfdfile ) {
 				printf("Error opening LITTLE ELF file: %s\n", elf_file_location);
 				// bfdfile = bfd_openr(elf_file_location, target );
-				bfd_perror( "Error on bfdfile" );	
+				bfd_perror( "Error on bfdfile" );
 				return(3);
 			}
 		}
-	
+
 		memset((void*) &DI, 0x00, sizeof(struct disassemble_info));
 
 		disassemble_fn = NULL;
-	
+
 		// important set up!
 		//---------------------------------------
 		ai.arch = arch;
 		ai.mach = mach;
 		bfd_set_arch_info(bfdfile, &ai);
 		//---------------------------------------
-	
+
 		/*
 		  bfd_error_type err = bfd_get_error();
 		  printf("bfd_error_msg: %s.\n", bfd_errmsg(err));
-		*/
+		 */
-	
+
+		/*if (disassemblerOptions[0] != '\0'){
+		    DI.disassembler_options = disassemblerOptions;
+		}*/
 		configureBfd(bfdfile, arch, mach, endian, &DI, &disassemble_fn);
 
 		/*
 		  err = bfd_get_error();
 		  printf("bfd_error_msg: %s.\n", bfd_errmsg(err));
-		*/
+		 */
-	
+
 		if (disassemble_fn == NULL){
 			fprintf(stderr, "Error: disassemble_fn is NULL. Nothing I can do.\n");
 			exit(1);
@@ -437,21 +447,25 @@ int main(int argc, char* argv[]){
 			/*
 			  printf("the disassemble_fn func pointer is: 0x%08X.\n", disassemble_fn);
 			  printf("We can try to disassemble for this arch/mach. calling disassemble_init_for_target().\n");
-			*/
+			 */
+
 			disassemble_init_for_target(&DI);
+			if (disassemblerOptions[0] != '\0'){
+				DI.disassembler_options = disassemblerOptions;
+			}
 
 			// go diassemble the buffer and build up the result in a accumulator string buffer.
-			processBuffer(byteBuffer, size >> 1, offset, disassemble_fn, &DI); // 
+			processBuffer(byteBuffer, size >> 1, offset, disassemble_fn, &DI); //
 
 		}
 
 		free((void*)bfdfile->xvec);
 		bfd_close(bfdfile);
-		
+
 		printf("EOF\n");
 		fflush(stdout);
 
 	} // while loop on lines of stdin
-	
+
 	return 0;
 }

Ghidra/Extensions/SleighDevTools/src/main/java/ghidra/app/util/disassemble/ExternalDisassembler.java

@@ -23,6 +23,8 @@ public interface ExternalDisassembler extends ExtensionPoint {
 
 	public String getDisassembly(CodeUnit cu) throws Exception;
 
+	public String getDisassemblyDisplayPrefix(CodeUnit cu) throws Exception;
+
 	public String getDisassemblyOfBytes(Language language, boolean isBigEndian, long address,
 			byte[] byteString) throws Exception;
 

Ghidra/Extensions/SleighDevTools/src/main/java/ghidra/app/util/disassemble/ExternalDisassemblyFieldFactory.java

@@ -150,7 +150,10 @@ public class ExternalDisassemblyFieldFactory extends FieldFactory {
 		if (disassembly == null) {
 			return null;
 		}
-
+		String prefix = disassembler.getDisassemblyDisplayPrefix(cu);
+		if (prefix != null) {
+			disassembly = prefix + " " + disassembly;
+		}
 		return disassembly;
 	}
 

Ghidra/Extensions/SleighDevTools/src/main/java/ghidra/app/util/disassemble/GNUExternalDisassembler.java

@@ -18,19 +18,23 @@ package ghidra.app.util.disassemble;
 import java.io.*;
 import java.util.*;
 
+import org.apache.commons.lang3.StringUtils;
+import org.jdom.*;
+import org.jdom.input.SAXBuilder;
+
 import generic.jar.ResourceFile;
 import ghidra.app.util.bin.ByteProvider;
 import ghidra.app.util.bin.MemoryByteProvider;
 import ghidra.framework.*;
 import ghidra.program.model.address.Address;
 import ghidra.program.model.address.AddressOutOfBoundsException;
-import ghidra.program.model.lang.Language;
+import ghidra.program.model.lang.*;
-import ghidra.program.model.lang.LanguageID;
+import ghidra.program.model.listing.*;
-import ghidra.program.model.listing.CodeUnit;
-import ghidra.program.model.listing.Program;
 import ghidra.program.model.mem.MemBuffer;
 import ghidra.program.model.mem.MemoryAccessException;
 import ghidra.util.Msg;
+import ghidra.util.xml.XmlUtilities;
+import util.CollectionUtils;
 
 public class GNUExternalDisassembler implements ExternalDisassembler {
 
@@ -39,6 +43,7 @@ public class GNUExternalDisassembler implements ExternalDisassembler {
 	// magic values for gdis that direct it to read bytes from stdin
 	private static final String READ_FROM_STDIN_PARAMETER = "stdin";
 	private static final String SEPARATOR_CHARACTER = "\n";
+	private static final String OPTIONS_SEPARATOR = "*";
 	private static final String ADDRESS_OUT_OF_BOUNDS = "is out of bounds.";
 	private static final String ENDING_STRING = "EOF";
 	private static final int NUM_BYTES = 32;
@@ -48,6 +53,8 @@ public class GNUExternalDisassembler implements ExternalDisassembler {
 	private static final String GDIS_EXE =
 		Platform.CURRENT_PLATFORM.getOperatingSystem() == OperatingSystem.WINDOWS ? "gdis.exe"
 				: "gdis";
+	private static final String EMPTY_DISASSEMBLER_OPTIONS = "";
+	private static final String GDIS_OPTIONS_FILENAME_PROPERTY = "gdis.disassembler.options.file";
 
 	private static HashMap<String, File> languageGdisMap;
 	private static File defaultGdisExecFile;
@@ -82,6 +89,21 @@ public class GNUExternalDisassembler implements ExternalDisassembler {
 		return gdisConfig != null && gdisConfig.architecture != UNSUPPORTED;
 	}
 
+	@Override
+	public String getDisassemblyDisplayPrefix(CodeUnit cu) throws Exception {
+		GdisConfig gdisConfig = checkLanguage(cu.getProgram().getLanguage());
+		if (gdisConfig == null || gdisConfig.architecture == UNSUPPORTED) {
+			return null;
+		}
+		Register contextRegister = gdisConfig.getContextRegister();
+		if (contextRegister == null) {
+			return null;
+		}
+		long value = getContextRegisterValue(cu, contextRegister);
+		String option = gdisConfig.getDisplayPrefixMap().get(value);
+		return option;
+	}
+
 	private static void reportMultipleMappings(Language language) {
 		List<String> externalNames = language.getLanguageDescription().getExternalNames("gnu");
 		if (externalNames != null && externalNames.size() > 1) {
@@ -89,8 +111,9 @@ public class GNUExternalDisassembler implements ExternalDisassembler {
 			StringBuilder sb = new StringBuilder();
 			boolean prependSeparator = false;
 			for (String name : externalNames) {
-				if (prependSeparator)
+				if (prependSeparator) {
 					sb.append(", ");
+				}
 				sb.append(name);
 				prependSeparator = true;
 			}
@@ -110,11 +133,17 @@ public class GNUExternalDisassembler implements ExternalDisassembler {
 		String machineId;
 		File gdisExecFile;
 		boolean usingDefault;
+		Register contextRegister;
+		Map<Long, String> valueToOptionString;
+		Map<Long, String> valueToDisplayPrefix;
+		Language lang;
+		String globalDisassemblerOptions;
 
 		GdisConfig(Language language, boolean isBigEndian) {
 
 			this.languageId = language.getLanguageID().toString();
 			this.isBigEndian = isBigEndian;
+			this.lang = language;
 
 			List<String> architectures = language.getLanguageDescription().getExternalNames("gnu");
 			//get first non-null
@@ -143,12 +172,123 @@ public class GNUExternalDisassembler implements ExternalDisassembler {
 				gdisExecFile = defaultGdisExecFile;
 				usingDefault = true;
 			}
+
+			List<String> gDisOptionsFile =
+				language.getLanguageDescription().getExternalNames(GDIS_OPTIONS_FILENAME_PROPERTY);
+			if (!CollectionUtils.isBlank(gDisOptionsFile)) {
+				try {
+					parseGdisOptionsFile(gDisOptionsFile.get(0));
+				}
+				catch (IOException e) {
+					Msg.error(this, "Error reading gdis options file " + e.getMessage());
+					contextRegister = null;
+					valueToOptionString = null;
+					valueToDisplayPrefix = null;
+				}
+			}
 		}
 
 		GdisConfig(Language lang) {
 			this(lang, lang.isBigEndian());
 		}
 
+		private void parseGdisOptionsFile(String fileName) throws IOException {
+			LanguageDescription desc = lang.getLanguageDescription();
+			if (!(desc instanceof SleighLanguageDescription)) {
+				throw new IOException("Not a Sleigh Language: " + lang.getLanguageID());
+			}
+
+			SleighLanguageDescription sld = (SleighLanguageDescription) desc;
+			ResourceFile defsFile = sld.getDefsFile();
+			ResourceFile parentFile = defsFile.getParentFile();
+			ResourceFile gdisOpts = new ResourceFile(parentFile, fileName);
+			SAXBuilder sax = XmlUtilities.createSecureSAXBuilder(false, false);
+			try (InputStream fis = gdisOpts.getInputStream()) {
+				Document doc = sax.build(fis);
+				Element rootElem = doc.getRootElement();
+				Element globalElement = rootElem.getChild("global");
+				if (globalElement != null) {
+					globalDisassemblerOptions = globalElement.getAttributeValue("optstring");
+				}
+				Element contextRegisterElement = rootElem.getChild("context_register");
+				if (contextRegisterElement == null) {
+					//no context_register element found in the xml file 
+					//this is not necessarily an error - might only be a global optstring
+					//global optstring has already been parsed, so we're done
+					if (globalElement != null) {
+						Msg.info(this,
+							"no context register element in " + gdisOpts.getAbsolutePath());
+						return;
+					}
+					//no context register element or global element, error
+					throw new JDOMException(
+						"No context_register element or global element in gdis options file");
+				}
+				if (contextRegisterElement.getContentSize() == 0) {
+					throw new JDOMException("No context register name provided.");
+				}
+				String contextRegisterName = contextRegisterElement.getContent(0).getValue();
+				contextRegister = lang.getRegister(contextRegisterName);
+				if (contextRegister == null) {
+					//the context register named in the xml file does not exist in the sleigh language
+					//this is an error
+					throw new JDOMException("Unknown context register " + contextRegisterName +
+						" for language " + lang.getLanguageID().getIdAsString());
+				}
+				valueToOptionString = new HashMap<>();
+				valueToDisplayPrefix = new HashMap<>();
+				Element options = rootElem.getChild("options");
+				List<Element> optList = options.getChildren("option");
+				for (Element opt : optList) {
+					Long value = Long.decode(opt.getAttributeValue("value"));
+					String optString = opt.getAttributeValue("optstring");
+					valueToOptionString.put(value, optString);
+					String displayPrefix = opt.getAttributeValue("display_prefix");
+					valueToDisplayPrefix.put(value, displayPrefix);
+				}
+
+			}
+			catch (JDOMException e) {
+				Msg.error(this, "Error reading " + fileName + ": " + e.getMessage());
+				contextRegister = null;
+				valueToOptionString = null;
+				valueToDisplayPrefix = null;
+			}
+		}
+
+		/**
+		 * Returns the global disassembler options
+		 * @return global option string, or {@code null} if the
+		 * global is unspecified.
+		 */
+		public String getGlobalDisassemblerOptions() {
+			return globalDisassemblerOptions;
+		}
+
+		/**
+		 * Return the context register determine by the gdis options file
+		 * @return the context register
+		 */
+		public Register getContextRegister() {
+			return contextRegister;
+		}
+
+		/**
+		 * Returns the map from context register values to gdis disassembler options
+		 * @return map values->options
+		 */
+		public Map<Long, String> getOptionsMap() {
+			return valueToOptionString;
+		}
+
+		/**
+		 * Returns the map from context register values to disassembly display prefixes
+		 * @return map values->prefixes
+		 */
+		public Map<Long, String> getDisplayPrefixMap() {
+			return valueToDisplayPrefix;
+		}
+
 		@Override
 		public boolean equals(Object obj) {
 			if (!(obj instanceof GdisConfig)) {
@@ -260,7 +400,7 @@ public class GNUExternalDisassembler implements ExternalDisassembler {
 
 		String bytes = getBytes(byteProvider, blockSize);
 
-		return runDisassembler(gdisConfig, address, bytes);
+		return runDisassembler(gdisConfig, address, bytes, EMPTY_DISASSEMBLER_OPTIONS);
 	}
 
 	public List<GnuDisassembledInstruction> getBlockDisassembly(Program program, Address addr,
@@ -272,8 +412,8 @@ public class GNUExternalDisassembler implements ExternalDisassembler {
 		int blockSize = pow2(blockSizeFactor);
 
 		Address blockAddr = addr.getNewAddress(addr.getOffset() & -blockSize); // block
-																				// aligned
+		// aligned
-																				// address
+		// address
 
 		return getBlockDisassembly(program.getLanguage(), blockAddr, blockSizeFactor,
 			new MemoryByteProvider(program.getMemory(), blockAddr));
@@ -301,7 +441,37 @@ public class GNUExternalDisassembler implements ExternalDisassembler {
 			return "";
 		}
 
-		List<GnuDisassembledInstruction> disassembly = runDisassembler(gdisConfig, address, bytes);
+		String disOptions = EMPTY_DISASSEMBLER_OPTIONS;
+		String globalOptions = gdisConfig.getGlobalDisassemblerOptions();
+		boolean hasGlobal = false;
+		if (globalOptions != null) {
+			hasGlobal = true;
+			//set the disOptions to the global options here in case there are global
+			//options but no options for context register values
+			disOptions = globalOptions;
+		}
+		Register contextRegister = gdisConfig.getContextRegister();
+		if (contextRegister != null) {
+			long value = getContextRegisterValue(cu, contextRegister);
+			String contextRegisterValueOption = gdisConfig.getOptionsMap().get(value);
+			if (contextRegisterValueOption == null) {
+				Msg.warn(this, "No option for value " + value + " of context register " +
+					contextRegister.getName());
+			}
+			else {
+				//need to put a comma between the global options and the options
+				//for this context register value
+				if (hasGlobal) {
+					disOptions = globalOptions + "," + contextRegisterValueOption;
+				}
+				//no global options, just send the options for this context register
+				else {
+					disOptions = contextRegisterValueOption;
+				}
+			}
+		}
+		List<GnuDisassembledInstruction> disassembly =
+			runDisassembler(gdisConfig, address, bytes, disOptions);
 
 		if (disassembly == null || disassembly.size() == 0 || disassembly.get(0) == null) {
 			return "(bad)";
@@ -324,7 +494,7 @@ public class GNUExternalDisassembler implements ExternalDisassembler {
 		String address = "0x" + Long.toHexString(addressOffset);
 
 		List<GnuDisassembledInstruction> disassembly =
-			runDisassembler(gdisConfig, address, bytesString);
+			runDisassembler(gdisConfig, address, bytesString, EMPTY_DISASSEMBLER_OPTIONS);
 
 		if (disassembly == null || disassembly.isEmpty() || disassembly.get(0) == null) {
 			return "(bad)";
@@ -332,14 +502,26 @@ public class GNUExternalDisassembler implements ExternalDisassembler {
 		return disassembly.get(0).toString();
 	}
 
+	private long getContextRegisterValue(CodeUnit cu, Register contextRegister) {
+		ProgramContext context = cu.getProgram().getProgramContext();
+		RegisterValue value = context.getRegisterValue(contextRegister, cu.getAddress());
+		if (value != null) {
+			return value.getUnsignedValue().longValue();
+		}
+		//we tried, return 0 to match SLEIGH default
+		return 0;
+	}
+
 	private String converBytesToString(byte[] bytes) {
 		String byteString = null;
 		for (byte thisByte : bytes) {
 			String thisByteString = Integer.toHexString(thisByte);
-			if (thisByteString.length() == 1)
+			if (thisByteString.length() == 1) {
 				thisByteString = "0" + thisByteString; // pad single digits
-			if (thisByteString.length() > 2)
+			}
+			if (thisByteString.length() > 2) {
 				thisByteString = thisByteString.substring(thisByteString.length() - 2);
+			}
 			// append this byte's hex string to the larger word length string
 			byteString = byteString + thisByteString;
 		}
@@ -402,7 +584,7 @@ public class GNUExternalDisassembler implements ExternalDisassembler {
 	}
 
 	private List<GnuDisassembledInstruction> runDisassembler(GdisConfig gdisConfig,
-			String addrString, String bytes) throws IOException {
+			String addrString, String bytes, String disassemblerOptions) throws IOException {
 
 		// if this is the first time running the disassembler process, or a
 		// parameter has changed (notably, not the address--we pass that in
@@ -414,8 +596,9 @@ public class GNUExternalDisassembler implements ExternalDisassembler {
 
 		if (disassemblerProcess == null || !sameConfig) {
 
-			if (!setupDisassembler(gdisConfig))
+			if (!setupDisassembler(gdisConfig)) {
 				return null;
+			}
 
 			outputWriter = new OutputStreamWriter(disassemblerProcess.getOutputStream());
 
@@ -432,7 +615,14 @@ public class GNUExternalDisassembler implements ExternalDisassembler {
 			return null; // if process previously died return nothing - quickly
 		}
 
-		String disassemblyRequest = addrString + SEPARATOR_CHARACTER + bytes + SEPARATOR_CHARACTER;
+		String disassemblyRequest = addrString + SEPARATOR_CHARACTER + bytes;
+		if (StringUtils.isEmpty(disassemblerOptions)) {
+			disassemblyRequest += '\n';
+		}
+		else {
+			disassemblyRequest += OPTIONS_SEPARATOR + disassemblerOptions + SEPARATOR_CHARACTER;
+		}
+
 		try {
 			outputWriter.write(disassemblyRequest);
 			outputWriter.flush();

Ghidra/Processors/ARM/certification.manifest

@@ -3,6 +3,7 @@ Module.manifest||GHIDRA||||END|
 build.gradle||GHIDRA||||END|
 data/languages/ARM.cspec||GHIDRA||||END|
 data/languages/ARM.dwarf||GHIDRA||||END|
+data/languages/ARM.gdis||GHIDRA||||END|
 data/languages/ARM.ldefs||GHIDRA||||END|
 data/languages/ARM.opinion||GHIDRA||||END|
 data/languages/ARM.sinc||GHIDRA||||END|

Ghidra/Processors/ARM/data/languages/ARM.gdis

@@ -0,0 +1,41 @@
+<!-- 
+     This file determines the disassembler options sent to the GNU external disassembler.
+     You can see which options are available for all architectures in the objdump man page.
+     Given a version of objdump compiled for a specific architecture, you can see what options
+     are available with "objdump -i -m" 
+     
+      The options for ARM are:
+       "reg-names-std" (the default)
+       "reg-names-apcs"
+       "reg-names-raw"
+       "reg-names-apcs"
+       "reg-names-special-apcs"
+       "force-thumb"      (force thumb disassembly)
+       "no-force-thumb"   (force arm disassembly)
+       (see the objdump manpage for the precise details about the different register 
+       naming options)
+     
+     "optstring" is the string of options, and "display_prefix" is an optional prefix prepended to the 
+     external disassembly field.  In this file, "A: " is prepended to ARM disassembly, and
+     "T: " is prepended to Thumb disassembly. 
+     
+     To send multiple options, concatenate them into a CSV list.  For example, "no-force-thumb,reg-names-raw"
+     is a valid optstring. You can also have a "global" element whose "optstring" attribute is 
+     always sent to the GNU disassembler.
+    
+     If there is not a context register defined, the global optstring is sent by itself (see x86-16.gdis for
+     an example).  If there is a context register defined, the global optstring is prepended to the 
+     optstring determined by a context register value.  
+     
+     This file must be listed in the .ldefs file for the processor, e.g.
+     <external_name tool="gdis.disassembler.options.file" name="ARM.gdis"/>
+-->
+
+<gdis> 
+    <context_register>TMode</context_register>
+    <options>
+        <option value="0x0" optstring="no-force-thumb" display_prefix = "A: "/>
+        <option value="0x1" optstring="force-thumb" display_prefix = "T: "/>
+    </options>
+</gdis>
+

Ghidra/Processors/ARM/data/languages/ARM.ldefs

@@ -14,6 +14,7 @@
     <compiler name="default" spec="ARM.cspec" id="default"/>
     <compiler name="Visual Studio" spec="ARM_win.cspec" id="windows"/>
     <external_name tool="gnu" name="iwmmxt"/>
+    <external_name tool="gdis.disassembler.options.file" name="ARM.gdis"/>
     <external_name tool="IDA-PRO" name="arm"/>
     <external_name tool="DWARF.register.mapping.file" name="ARMneon.dwarf"/>
   </language>
@@ -31,6 +32,7 @@
     <compiler name="default" spec="ARM.cspec" id="default"/>
     <compiler name="Visual Studio" spec="ARM_win.cspec" id="windows"/>
     <external_name tool="gnu" name="iwmmxt"/>
+    <external_name tool="gdis.disassembler.options.file" name="ARM.gdis"/>
     <external_name tool="IDA-PRO" name="arm"/>
     <external_name tool="DWARF.register.mapping.file" name="ARMneon.dwarf"/>
   </language>
@@ -63,6 +65,7 @@
     <description>Generic ARM/Thumb v8 big endian</description>
     <compiler name="default" spec="ARM.cspec" id="default"/>
     <external_name tool="gnu" name="iwmmxt"/>
+    <external_name tool="gdis.disassembler.options.file" name="ARM.gdis"/>
     <external_name tool="IDA-PRO" name="armb"/>
     <external_name tool="DWARF.register.mapping.file" name="ARMneon.dwarf"/>
   </language>
@@ -96,6 +99,7 @@
     <compiler name="default" spec="ARM.cspec" id="default"/>
     <compiler name="Visual Studio" spec="ARM_win.cspec" id="windows"/>
     <external_name tool="gnu" name="iwmmxt"/>
+    <external_name tool="gdis.disassembler.options.file" name="ARM.gdis"/>
     <external_name tool="IDA-PRO" name="arm"/>
     <external_name tool="DWARF.register.mapping.file" name="ARMneon.dwarf"/>
   </language>
@@ -128,6 +132,7 @@
     <description>Generic ARM/Thumb v7 big endian</description>
     <compiler name="default" spec="ARM.cspec" id="default"/>
     <external_name tool="gnu" name="iwmmxt"/>
+    <external_name tool="gdis.disassembler.options.file" name="ARM.gdis"/>
     <external_name tool="IDA-PRO" name="armb"/>
     <external_name tool="DWARF.register.mapping.file" name="ARMneon.dwarf"/>
   </language>
@@ -174,6 +179,7 @@
     <description>Generic ARM/Thumb v6 little endian</description>
     <compiler name="default" spec="ARM_v45.cspec" id="default"/>
     <external_name tool="gnu" name="xscale"/>
+    <external_name tool="gdis.disassembler.options.file" name="ARM.gdis"/>
     <external_name tool="IDA-PRO" name="arm"/>
     <external_name tool="DWARF.register.mapping.file" name="ARM.dwarf"/>
     <!-- change DWARF register mapping to ARMneon.dwarf if VFPv2 is enabled --> 
@@ -192,6 +198,7 @@
     <description>Generic ARM/Thumb v6 big endian</description>
     <compiler name="default" spec="ARM_v45.cspec" id="default"/>
     <external_name tool="gnu" name="xscale"/>
+    <external_name tool="gdis.disassembler.options.file" name="ARM.gdis"/>
     <external_name tool="IDA-PRO" name="armb"/>
     <external_name tool="DWARF.register.mapping.file" name="ARM.dwarf"/>
     <!-- change DWARF register mapping to ARMneon.dwarf if VFPv2 is enabled --> 
@@ -210,6 +217,7 @@
     <description>Generic ARM/Thumb v5 little endian (T-variant)</description>
     <compiler name="default" spec="ARM_v45.cspec" id="default"/>
     <external_name tool="gnu" name="armv5t"/>
+    <external_name tool="gdis.disassembler.options.file" name="ARM.gdis"/>
     <external_name tool="IDA-PRO" name="arm"/>
     <external_name tool="DWARF.register.mapping.file" name="ARM.dwarf"/>
   </language>
@@ -226,6 +234,7 @@
     <description>Generic ARM/Thumb v5 big endian (T-variant)</description>
     <compiler name="default" spec="ARM_v45.cspec" id="default"/>
     <external_name tool="gnu" name="armv5t"/>
+    <external_name tool="gdis.disassembler.options.file" name="ARM.gdis"/>
     <external_name tool="IDA-PRO" name="armb"/>
     <external_name tool="DWARF.register.mapping.file" name="ARM.dwarf"/>
   </language>
@@ -274,6 +283,7 @@
     <description>Generic ARM/Thumb v4 little endian (T-variant)</description>
     <compiler name="default" spec="ARM_v45.cspec" id="default"/>
     <external_name tool="gnu" name="armv4t"/>
+    <external_name tool="gdis.disassembler.options.file" name="ARM.gdis"/>
     <external_name tool="IDA-PRO" name="arm"/>
     <external_name tool="DWARF.register.mapping.file" name="ARM.dwarf"/>
   </language>
@@ -290,6 +300,7 @@
     <description>Generic ARM/Thumb v4 big endian (T-variant)</description>
     <compiler name="default" spec="ARM_v45.cspec" id="default"/>
     <external_name tool="gnu" name="armv4t"/>
+    <external_name tool="gdis.disassembler.options.file" name="ARM.gdis"/>
     <external_name tool="IDA-PRO" name="armb"/>
     <external_name tool="DWARF.register.mapping.file" name="ARM.dwarf"/>
   </language>

Ghidra/Processors/x86/certification.manifest

@@ -30,6 +30,7 @@ data/languages/sha.sinc||GHIDRA||||END|
 data/languages/smx.sinc||GHIDRA||||END|
 data/languages/x86-16-real.pspec||GHIDRA||||END|
 data/languages/x86-16.cspec||GHIDRA||||END|
+data/languages/x86-16.gdis||GHIDRA||||END|
 data/languages/x86-16.pspec||GHIDRA||||END|
 data/languages/x86-64-gcc.cspec||GHIDRA||||END|
 data/languages/x86-64-win.cspec||GHIDRA||||END|

Ghidra/Processors/x86/data/languages/x86-16.gdis

@@ -0,0 +1,3 @@
+<gdis>
+    <global optstring="intel"/>
+</gdis>

Ghidra/Processors/x86/data/languages/x86.ldefs

@@ -59,6 +59,8 @@
 	<external_name tool="IDA-PRO" name="80486r"/>
 	<external_name tool="IDA-PRO" name="80586r"/>
 	<external_name tool="IDA-PRO" name="metapc"/>
+    <external_name tool="gnu" name="i8086"/>
+    <external_name tool="gdis.disassembler.options.file" name="x86-16.gdis"/>
     </language>
   <language processor="x86"
             endian="little"
@@ -72,6 +74,8 @@
     <description>Intel/AMD 16-bit x86 Protected Mode</description>
     <compiler name="default" spec="x86-16.cspec" id="default"/>
 	<external_name tool="IDA-PRO" name="8086p"/>
+    <external_name tool="gnu" name="i8086"/>
+    <external_name tool="gdis.disassembler.options.file" name="x86-16.gdis"/>
     </language>
     <language processor="x86"
             endian="little"