mirrors/bird
mirrors/bird
0
0
Fork 0
mirror of https://gitlab.nic.cz/labs/bird.git synced 2024-11-13 22:58:42 +00:00
Code Releases Activity
mq-config-bg
bird/filter/f-util.c

110 lines
2.6 KiB
C
Raw Permalink Normal View History

Filters, second try. This time they have their own directory.
1999-01-15 16:49:17 +00:00
/*
* Filters: utility functions
*
* Copyright 1998 Pavel Machek <pavel@ucw.cz>
Filter: Instruction codes named as enum The two-letter instructions were quite messy but they could be easily read from memory dumps. Now GDB (since 2012) supports pretty printing enum values and GCC checks the switch construction for missing enum values so we are converting the nice two-byte values to enums. Anyway, the enum still keeps the old two-byte values to be able to read the instruction codes even without GDB from plain memory dump.
2017-10-19 10:39:44 +00:00
* 2017 Jan Maria Matejka <mq@ucw.cz>
Filters, second try. This time they have their own directory.
1999-01-15 16:49:17 +00:00
*
* Can be freely distributed and used under the terms of the GNU GPL.
*/
#include "nest/bird.h"
#include "conf/conf.h"
Config: Make the parser and lexer reentrant. This is part of the multithreading journey. The parser and lexer were using loads of global variables and all of these are now packed into struct cf_context and others. Note that the config API has changed: * cfg_alloc[zu]?(size) is now cf_alloc[zu]?(ctx, size) * cf_error(msg, ...) is now cf_error(ctx, msg, ...) * config_parse() and cli_parse() are now called differently * there is a brand new CF_CTX section in *.Y files which participates in struct cf_context construction
2018-08-13 09:41:27 +00:00
#include "conf/parser.h"
Filters, second try. This time they have their own directory.
1999-01-15 16:49:17 +00:00
#include "filter/filter.h"
Config: Make the parser and lexer reentrant. This is part of the multithreading journey. The parser and lexer were using loads of global variables and all of these are now packed into struct cf_context and others. Note that the config API has changed: * cfg_alloc[zu]?(size) is now cf_alloc[zu]?(ctx, size) * cf_error(msg, ...) is now cf_error(ctx, msg, ...) * config_parse() and cli_parse() are now called differently * there is a brand new CF_CTX section in *.Y files which participates in struct cf_context construction
2018-08-13 09:41:27 +00:00
#include "filter/f-util.h"
Filters, second try. This time they have their own directory.
1999-01-15 16:49:17 +00:00
Create syntax sugar for add/delete/prepend, so xyzzy.prepend(123) is possible. That means that milestone 3 was reached.
2000-04-20 10:25:51 +00:00
#define P(a,b) ((a<<8) | b)
Filters are now a tiny bit stronger (if is actually working ;-)
1999-03-08 20:30:06 +00:00
struct f_inst *
Config: Make the parser and lexer reentrant. This is part of the multithreading journey. The parser and lexer were using loads of global variables and all of these are now packed into struct cf_context and others. Note that the config API has changed: * cfg_alloc[zu]?(size) is now cf_alloc[zu]?(ctx, size) * cf_error(msg, ...) is now cf_error(ctx, msg, ...) * config_parse() and cli_parse() are now called differently * there is a brand new CF_CTX section in *.Y files which participates in struct cf_context construction
2018-08-13 09:41:27 +00:00
f_new_inst(struct cf_context *ctx, enum f_instruction_code fi_code)
Add interface for running filters (please comment!), avoid bison warnings
1999-03-02 19:49:28 +00:00
{
Filters are now a tiny bit stronger (if is actually working ;-)
1999-03-08 20:30:06 +00:00
struct f_inst * ret;
Filter: Instruction codes named as enum The two-letter instructions were quite messy but they could be easily read from memory dumps. Now GDB (since 2012) supports pretty printing enum values and GCC checks the switch construction for missing enum values so we are converting the nice two-byte values to enums. Anyway, the enum still keeps the old two-byte values to be able to read the instruction codes even without GDB from plain memory dump.
2017-10-19 10:39:44 +00:00
ret = cfg_allocz(sizeof(struct f_inst));
ret->fi_code = fi_code;
Config: Make the parser and lexer reentrant. This is part of the multithreading journey. The parser and lexer were using loads of global variables and all of these are now packed into struct cf_context and others. Note that the config API has changed: * cfg_alloc[zu]?(size) is now cf_alloc[zu]?(ctx, size) * cf_error(msg, ...) is now cf_error(ctx, msg, ...) * config_parse() and cli_parse() are now called differently * there is a brand new CF_CTX section in *.Y files which participates in struct cf_context construction
2018-08-13 09:41:27 +00:00
ret->lineno = ctx->order->state->lino;
Add interface for running filters (please comment!), avoid bison warnings
1999-03-02 19:49:28 +00:00
return ret;
}
Dynamic attributes are now declared in per-protocol grammar files instead of filter/config.Y. Bird now compiles even if you disable RIP. Removed RTA and IMPOSSIBLE tokens (unused). Removed superfluous comment in filter.h. I've tried to do my best, but Pavel, please check these changes.
2000-03-01 11:32:23 +00:00
struct f_inst *
Config: Make the parser and lexer reentrant. This is part of the multithreading journey. The parser and lexer were using loads of global variables and all of these are now packed into struct cf_context and others. Note that the config API has changed: * cfg_alloc[zu]?(size) is now cf_alloc[zu]?(ctx, size) * cf_error(msg, ...) is now cf_error(ctx, msg, ...) * config_parse() and cli_parse() are now called differently * there is a brand new CF_CTX section in *.Y files which participates in struct cf_context construction
2018-08-13 09:41:27 +00:00
f_new_inst_da(struct cf_context *ctx, enum f_instruction_code fi_code, struct f_dynamic_attr da)
Dynamic attributes are now declared in per-protocol grammar files instead of filter/config.Y. Bird now compiles even if you disable RIP. Removed RTA and IMPOSSIBLE tokens (unused). Removed superfluous comment in filter.h. I've tried to do my best, but Pavel, please check these changes.
2000-03-01 11:32:23 +00:00
{
Config: Make the parser and lexer reentrant. This is part of the multithreading journey. The parser and lexer were using loads of global variables and all of these are now packed into struct cf_context and others. Note that the config API has changed: * cfg_alloc[zu]?(size) is now cf_alloc[zu]?(ctx, size) * cf_error(msg, ...) is now cf_error(ctx, msg, ...) * config_parse() and cli_parse() are now called differently * there is a brand new CF_CTX section in *.Y files which participates in struct cf_context construction
2018-08-13 09:41:27 +00:00
struct f_inst *ret = f_new_inst(ctx, fi_code);
Merge branch 'master' into int-new
2018-03-13 15:51:04 +00:00
ret->aux = (da.f_type << 8) | da.type;
Filter: Instruction codes named as enum The two-letter instructions were quite messy but they could be easily read from memory dumps. Now GDB (since 2012) supports pretty printing enum values and GCC checks the switch construction for missing enum values so we are converting the nice two-byte values to enums. Anyway, the enum still keeps the old two-byte values to be able to read the instruction codes even without GDB from plain memory dump.
2017-10-19 10:39:44 +00:00
ret->a2.i = da.ea_code;
return ret;
}
struct f_inst *
Config: Make the parser and lexer reentrant. This is part of the multithreading journey. The parser and lexer were using loads of global variables and all of these are now packed into struct cf_context and others. Note that the config API has changed: * cfg_alloc[zu]?(size) is now cf_alloc[zu]?(ctx, size) * cf_error(msg, ...) is now cf_error(ctx, msg, ...) * config_parse() and cli_parse() are now called differently * there is a brand new CF_CTX section in *.Y files which participates in struct cf_context construction
2018-08-13 09:41:27 +00:00
f_new_inst_sa(struct cf_context *ctx, enum f_instruction_code fi_code, struct f_static_attr sa)
Filter: Instruction codes named as enum The two-letter instructions were quite messy but they could be easily read from memory dumps. Now GDB (since 2012) supports pretty printing enum values and GCC checks the switch construction for missing enum values so we are converting the nice two-byte values to enums. Anyway, the enum still keeps the old two-byte values to be able to read the instruction codes even without GDB from plain memory dump.
2017-10-19 10:39:44 +00:00
{
Config: Make the parser and lexer reentrant. This is part of the multithreading journey. The parser and lexer were using loads of global variables and all of these are now packed into struct cf_context and others. Note that the config API has changed: * cfg_alloc[zu]?(size) is now cf_alloc[zu]?(ctx, size) * cf_error(msg, ...) is now cf_error(ctx, msg, ...) * config_parse() and cli_parse() are now called differently * there is a brand new CF_CTX section in *.Y files which participates in struct cf_context construction
2018-08-13 09:41:27 +00:00
struct f_inst *ret = f_new_inst(ctx, fi_code);
Filter: Instruction codes named as enum The two-letter instructions were quite messy but they could be easily read from memory dumps. Now GDB (since 2012) supports pretty printing enum values and GCC checks the switch construction for missing enum values so we are converting the nice two-byte values to enums. Anyway, the enum still keeps the old two-byte values to be able to read the instruction codes even without GDB from plain memory dump.
2017-10-19 10:39:44 +00:00
ret->aux = sa.f_type;
ret->a2.i = sa.sa_code;
ret->a1.i = sa.readonly;
return ret;
Dynamic attributes are now declared in per-protocol grammar files instead of filter/config.Y. Bird now compiles even if you disable RIP. Removed RTA and IMPOSSIBLE tokens (unused). Removed superfluous comment in filter.h. I've tried to do my best, but Pavel, please check these changes.
2000-03-01 11:32:23 +00:00
}
Create syntax sugar for add/delete/prepend, so xyzzy.prepend(123) is possible. That means that milestone 3 was reached.
2000-04-20 10:25:51 +00:00
/*
* Generate set_dynamic( operation( get_dynamic(), argument ) )
*/
struct f_inst *
Config: Make the parser and lexer reentrant. This is part of the multithreading journey. The parser and lexer were using loads of global variables and all of these are now packed into struct cf_context and others. Note that the config API has changed: * cfg_alloc[zu]?(size) is now cf_alloc[zu]?(ctx, size) * cf_error(msg, ...) is now cf_error(ctx, msg, ...) * config_parse() and cli_parse() are now called differently * there is a brand new CF_CTX section in *.Y files which participates in struct cf_context construction
2018-08-13 09:41:27 +00:00
f_generate_complex(struct cf_context *ctx, int operation, int operation_aux, struct f_dynamic_attr da, struct f_inst *argument)
Create syntax sugar for add/delete/prepend, so xyzzy.prepend(123) is possible. That means that milestone 3 was reached.
2000-04-20 10:25:51 +00:00
{
Config: Make the parser and lexer reentrant. This is part of the multithreading journey. The parser and lexer were using loads of global variables and all of these are now packed into struct cf_context and others. Note that the config API has changed: * cfg_alloc[zu]?(size) is now cf_alloc[zu]?(ctx, size) * cf_error(msg, ...) is now cf_error(ctx, msg, ...) * config_parse() and cli_parse() are now called differently * there is a brand new CF_CTX section in *.Y files which participates in struct cf_context construction
2018-08-13 09:41:27 +00:00
struct f_inst *set_dyn = f_new_inst_da(ctx, FI_EA_SET, da),
*oper = f_new_inst(ctx, operation),
*get_dyn = f_new_inst_da(ctx, FI_EA_GET, da);
Create syntax sugar for add/delete/prepend, so xyzzy.prepend(123) is possible. That means that milestone 3 was reached.
2000-04-20 10:25:51 +00:00
oper->aux = operation_aux;
oper->a1.p = get_dyn;
oper->a2.p = argument;
Filter: Instruction codes named as enum The two-letter instructions were quite messy but they could be easily read from memory dumps. Now GDB (since 2012) supports pretty printing enum values and GCC checks the switch construction for missing enum values so we are converting the nice two-byte values to enums. Anyway, the enum still keeps the old two-byte values to be able to read the instruction codes even without GDB from plain memory dump.
2017-10-19 10:39:44 +00:00
Create syntax sugar for add/delete/prepend, so xyzzy.prepend(123) is possible. That means that milestone 3 was reached.
2000-04-20 10:25:51 +00:00
set_dyn->a1.p = oper;
return set_dyn;
}
Route Origin Authorization basics. - ROA tables, which are used as a basic part for RPKI. - Commands for examining and modifying ROA tables. - Filter operators based on ROA tables consistent with RFC 6483.
2012-03-18 16:32:30 +00:00
struct f_inst *
Config: Make the parser and lexer reentrant. This is part of the multithreading journey. The parser and lexer were using loads of global variables and all of these are now packed into struct cf_context and others. Note that the config API has changed: * cfg_alloc[zu]?(size) is now cf_alloc[zu]?(ctx, size) * cf_error(msg, ...) is now cf_error(ctx, msg, ...) * config_parse() and cli_parse() are now called differently * there is a brand new CF_CTX section in *.Y files which participates in struct cf_context construction
2018-08-13 09:41:27 +00:00
f_generate_roa_check(struct cf_context *ctx, struct rtable_config *table, struct f_inst *prefix, struct f_inst *asn)
Route Origin Authorization basics. - ROA tables, which are used as a basic part for RPKI. - Commands for examining and modifying ROA tables. - Filter operators based on ROA tables consistent with RFC 6483.
2012-03-18 16:32:30 +00:00
{
struct f_inst_roa_check *ret = cfg_allocz(sizeof(struct f_inst_roa_check));
Filter: Instruction codes named as enum The two-letter instructions were quite messy but they could be easily read from memory dumps. Now GDB (since 2012) supports pretty printing enum values and GCC checks the switch construction for missing enum values so we are converting the nice two-byte values to enums. Anyway, the enum still keeps the old two-byte values to be able to read the instruction codes even without GDB from plain memory dump.
2017-10-19 10:39:44 +00:00
ret->i.fi_code = FI_ROA_CHECK;
Config: Make the parser and lexer reentrant. This is part of the multithreading journey. The parser and lexer were using loads of global variables and all of these are now packed into struct cf_context and others. Note that the config API has changed: * cfg_alloc[zu]?(size) is now cf_alloc[zu]?(ctx, size) * cf_error(msg, ...) is now cf_error(ctx, msg, ...) * config_parse() and cli_parse() are now called differently * there is a brand new CF_CTX section in *.Y files which participates in struct cf_context construction
2018-08-13 09:41:27 +00:00
ret->i.lineno = ctx->order->state->lino;
Route Origin Authorization basics. - ROA tables, which are used as a basic part for RPKI. - Commands for examining and modifying ROA tables. - Filter operators based on ROA tables consistent with RFC 6483.
2012-03-18 16:32:30 +00:00
ret->i.arg1 = prefix;
ret->i.arg2 = asn;
/* prefix == NULL <-> asn == NULL */
Rewrite roa_check() for integrated BIRD Thanks to Ondrej Zajicek for his support with writing this code.
2016-01-20 14:38:37 +00:00
if (table->addr_type != NET_ROA4 && table->addr_type != NET_ROA6)
Config: Make the parser and lexer reentrant. This is part of the multithreading journey. The parser and lexer were using loads of global variables and all of these are now packed into struct cf_context and others. Note that the config API has changed: * cfg_alloc[zu]?(size) is now cf_alloc[zu]?(ctx, size) * cf_error(msg, ...) is now cf_error(ctx, msg, ...) * config_parse() and cli_parse() are now called differently * there is a brand new CF_CTX section in *.Y files which participates in struct cf_context construction
2018-08-13 09:41:27 +00:00
cf_error(ctx, "%s is not a ROA table", table->name);
Rewrite roa_check() for integrated BIRD Thanks to Ondrej Zajicek for his support with writing this code.
2016-01-20 14:38:37 +00:00
ret->rtc = table;
Route Origin Authorization basics. - ROA tables, which are used as a basic part for RPKI. - Commands for examining and modifying ROA tables. - Filter operators based on ROA tables consistent with RFC 6483.
2012-03-18 16:32:30 +00:00
return &ret->i;
}
Filter: instruction names
2018-04-30 10:49:22 +00:00
static const char * const f_instruction_name_str[] = {
#define F(c,a,b) \
[c] = #c,
FI__LIST
#undef F
};
const char *
f_instruction_name(enum f_instruction_code fi)
{
if (fi < FI__MAX)
return f_instruction_name_str[fi];
else
bug("Got unknown instruction code: %d", fi);
}
Several filter changes. (Pavel, please check if they are OK.) o Changed parameters of f_run. Changed rtein+rteout pair to rte pointer passed by reference, added ea_list of temporary attrs again passed by reference and finally added a pointer to memory pool for storing temporary data (new ea_lists's, temporary rta's etc.). o Re-ordered result codes, so that all accepts come before all rejects. o Introduced FILTER_ACCEPT and FILTER_REJECT dummy values (will be used in protocol configurations). o Added filter_name() which returns name of a filter or ACCEPT/REJECT for the dummies.
1999-04-05 20:10:31 +00:00
char *
filter_name(struct filter *filter)
{
if (!filter)
return "ACCEPT";
else if (filter == FILTER_REJECT)
return "REJECT";
Fixes name for unnamed filters. Thanks to Alexander V. Chernikov for the suggestion.
2012-03-15 11:50:49 +00:00
else if (!filter->name)
return "(unnamed)";
Several filter changes. (Pavel, please check if they are OK.) o Changed parameters of f_run. Changed rtein+rteout pair to rte pointer passed by reference, added ea_list of temporary attrs again passed by reference and finally added a pointer to memory pool for storing temporary data (new ea_lists's, temporary rta's etc.). o Re-ordered result codes, so that all accepts come before all rejects. o Introduced FILTER_ACCEPT and FILTER_REJECT dummy values (will be used in protocol configurations). o Added filter_name() which returns name of a filter or ACCEPT/REJECT for the dummies.
1999-04-05 20:10:31 +00:00
else
return filter->name;
}
Copy Permalink