auto import from //branches/cupcake/...@132276
diff --git a/libc/SYSCALLS.TXT b/libc/SYSCALLS.TXT
index 0445fc9..1ef06d2 100644
--- a/libc/SYSCALLS.TXT
+++ b/libc/SYSCALLS.TXT
@@ -216,22 +216,6 @@
int sched_get_priority_min(int policy) 160
int sched_rr_get_interval(pid_t pid, struct timespec *interval) 161
-# system-V inter-process communication
-# TODO: implement x86 stubs for these functions when needed (when ?)
-#
-int semctl(int semid, int semnum, int cmd, ...) 300,-1
-int semget(key_t key, int nsems, int semflg) 299,-1
-int semop(int semid, struct sembuf* sops, size_t nsops) 298,-1
-void* shmat(int shmid, const void* shmaddr, int shmflg) 305,-1
-int shmctl(int shmid, int cmd, struct shmid_ds* buf) 308,-1
-int shmdt(const void* shmaddr) 306,-1
-int shmget(key_t key, size_t size, int shmflg) 307,-1
-int msgctl(int msqid, int cmd, struct msqid_ds *buf) 304,-1
-int msgget(key_t key, int msgflg) 303,-1
-int msgrcv(int msqid, void* msgp, size_t msgsz, long int msgtyp, int msgflg) 302,-1
-int msgsnd(int msqid, const void* msgp, size_t msgsz, int msgflg) 301,-1
-
-
# other
int uname(struct utsname *) 122
pid_t __wait4:wait4(pid_t pid, int *status, int options, struct rusage *rusage) 114
diff --git a/libc/arch-arm/syscalls.mk b/libc/arch-arm/syscalls.mk
index a140a59..706cb0c 100644
--- a/libc/arch-arm/syscalls.mk
+++ b/libc/arch-arm/syscalls.mk
@@ -150,17 +150,6 @@
syscall_src += arch-arm/syscalls/sched_get_priority_max.S
syscall_src += arch-arm/syscalls/sched_get_priority_min.S
syscall_src += arch-arm/syscalls/sched_rr_get_interval.S
-syscall_src += arch-arm/syscalls/semctl.S
-syscall_src += arch-arm/syscalls/semget.S
-syscall_src += arch-arm/syscalls/semop.S
-syscall_src += arch-arm/syscalls/shmat.S
-syscall_src += arch-arm/syscalls/shmctl.S
-syscall_src += arch-arm/syscalls/shmdt.S
-syscall_src += arch-arm/syscalls/shmget.S
-syscall_src += arch-arm/syscalls/msgctl.S
-syscall_src += arch-arm/syscalls/msgget.S
-syscall_src += arch-arm/syscalls/msgrcv.S
-syscall_src += arch-arm/syscalls/msgsnd.S
syscall_src += arch-arm/syscalls/uname.S
syscall_src += arch-arm/syscalls/__wait4.S
syscall_src += arch-arm/syscalls/umask.S
diff --git a/libc/arch-arm/syscalls/msgctl.S b/libc/arch-arm/syscalls/msgctl.S
deleted file mode 100644
index 5c0bbf5..0000000
--- a/libc/arch-arm/syscalls/msgctl.S
+++ /dev/null
@@ -1,19 +0,0 @@
-/* autogenerated by gensyscalls.py */
-#include <sys/linux-syscalls.h>
-
- .text
- .type msgctl, #function
- .globl msgctl
- .align 4
- .fnstart
-
-msgctl:
- .save {r4, r7}
- stmfd sp!, {r4, r7}
- ldr r7, =__NR_msgctl
- swi #0
- ldmfd sp!, {r4, r7}
- movs r0, r0
- bxpl lr
- b __set_syscall_errno
- .fnend
diff --git a/libc/arch-arm/syscalls/msgget.S b/libc/arch-arm/syscalls/msgget.S
deleted file mode 100644
index ac16440..0000000
--- a/libc/arch-arm/syscalls/msgget.S
+++ /dev/null
@@ -1,19 +0,0 @@
-/* autogenerated by gensyscalls.py */
-#include <sys/linux-syscalls.h>
-
- .text
- .type msgget, #function
- .globl msgget
- .align 4
- .fnstart
-
-msgget:
- .save {r4, r7}
- stmfd sp!, {r4, r7}
- ldr r7, =__NR_msgget
- swi #0
- ldmfd sp!, {r4, r7}
- movs r0, r0
- bxpl lr
- b __set_syscall_errno
- .fnend
diff --git a/libc/arch-arm/syscalls/msgrcv.S b/libc/arch-arm/syscalls/msgrcv.S
deleted file mode 100644
index 8db4125..0000000
--- a/libc/arch-arm/syscalls/msgrcv.S
+++ /dev/null
@@ -1,21 +0,0 @@
-/* autogenerated by gensyscalls.py */
-#include <sys/linux-syscalls.h>
-
- .text
- .type msgrcv, #function
- .globl msgrcv
- .align 4
- .fnstart
-
-msgrcv:
- mov ip, sp
- .save {r4, r5, r6, r7}
- stmfd sp!, {r4, r5, r6, r7}
- ldmfd ip, {r4, r5, r6}
- ldr r7, =__NR_msgrcv
- swi #0
- ldmfd sp!, {r4, r5, r6, r7}
- movs r0, r0
- bxpl lr
- b __set_syscall_errno
- .fnend
diff --git a/libc/arch-arm/syscalls/msgsnd.S b/libc/arch-arm/syscalls/msgsnd.S
deleted file mode 100644
index 02c92ec..0000000
--- a/libc/arch-arm/syscalls/msgsnd.S
+++ /dev/null
@@ -1,19 +0,0 @@
-/* autogenerated by gensyscalls.py */
-#include <sys/linux-syscalls.h>
-
- .text
- .type msgsnd, #function
- .globl msgsnd
- .align 4
- .fnstart
-
-msgsnd:
- .save {r4, r7}
- stmfd sp!, {r4, r7}
- ldr r7, =__NR_msgsnd
- swi #0
- ldmfd sp!, {r4, r7}
- movs r0, r0
- bxpl lr
- b __set_syscall_errno
- .fnend
diff --git a/libc/arch-arm/syscalls/semctl.S b/libc/arch-arm/syscalls/semctl.S
deleted file mode 100644
index a50e4dc..0000000
--- a/libc/arch-arm/syscalls/semctl.S
+++ /dev/null
@@ -1,19 +0,0 @@
-/* autogenerated by gensyscalls.py */
-#include <sys/linux-syscalls.h>
-
- .text
- .type semctl, #function
- .globl semctl
- .align 4
- .fnstart
-
-semctl:
- .save {r4, r7}
- stmfd sp!, {r4, r7}
- ldr r7, =__NR_semctl
- swi #0
- ldmfd sp!, {r4, r7}
- movs r0, r0
- bxpl lr
- b __set_syscall_errno
- .fnend
diff --git a/libc/arch-arm/syscalls/semget.S b/libc/arch-arm/syscalls/semget.S
deleted file mode 100644
index aaabd37..0000000
--- a/libc/arch-arm/syscalls/semget.S
+++ /dev/null
@@ -1,19 +0,0 @@
-/* autogenerated by gensyscalls.py */
-#include <sys/linux-syscalls.h>
-
- .text
- .type semget, #function
- .globl semget
- .align 4
- .fnstart
-
-semget:
- .save {r4, r7}
- stmfd sp!, {r4, r7}
- ldr r7, =__NR_semget
- swi #0
- ldmfd sp!, {r4, r7}
- movs r0, r0
- bxpl lr
- b __set_syscall_errno
- .fnend
diff --git a/libc/arch-arm/syscalls/semop.S b/libc/arch-arm/syscalls/semop.S
deleted file mode 100644
index 20ef68b..0000000
--- a/libc/arch-arm/syscalls/semop.S
+++ /dev/null
@@ -1,19 +0,0 @@
-/* autogenerated by gensyscalls.py */
-#include <sys/linux-syscalls.h>
-
- .text
- .type semop, #function
- .globl semop
- .align 4
- .fnstart
-
-semop:
- .save {r4, r7}
- stmfd sp!, {r4, r7}
- ldr r7, =__NR_semop
- swi #0
- ldmfd sp!, {r4, r7}
- movs r0, r0
- bxpl lr
- b __set_syscall_errno
- .fnend
diff --git a/libc/arch-arm/syscalls/shmat.S b/libc/arch-arm/syscalls/shmat.S
deleted file mode 100644
index 8b63bf0..0000000
--- a/libc/arch-arm/syscalls/shmat.S
+++ /dev/null
@@ -1,19 +0,0 @@
-/* autogenerated by gensyscalls.py */
-#include <sys/linux-syscalls.h>
-
- .text
- .type shmat, #function
- .globl shmat
- .align 4
- .fnstart
-
-shmat:
- .save {r4, r7}
- stmfd sp!, {r4, r7}
- ldr r7, =__NR_shmat
- swi #0
- ldmfd sp!, {r4, r7}
- movs r0, r0
- bxpl lr
- b __set_syscall_errno
- .fnend
diff --git a/libc/arch-arm/syscalls/shmctl.S b/libc/arch-arm/syscalls/shmctl.S
deleted file mode 100644
index a29caa6..0000000
--- a/libc/arch-arm/syscalls/shmctl.S
+++ /dev/null
@@ -1,19 +0,0 @@
-/* autogenerated by gensyscalls.py */
-#include <sys/linux-syscalls.h>
-
- .text
- .type shmctl, #function
- .globl shmctl
- .align 4
- .fnstart
-
-shmctl:
- .save {r4, r7}
- stmfd sp!, {r4, r7}
- ldr r7, =__NR_shmctl
- swi #0
- ldmfd sp!, {r4, r7}
- movs r0, r0
- bxpl lr
- b __set_syscall_errno
- .fnend
diff --git a/libc/arch-arm/syscalls/shmdt.S b/libc/arch-arm/syscalls/shmdt.S
deleted file mode 100644
index bee707e..0000000
--- a/libc/arch-arm/syscalls/shmdt.S
+++ /dev/null
@@ -1,19 +0,0 @@
-/* autogenerated by gensyscalls.py */
-#include <sys/linux-syscalls.h>
-
- .text
- .type shmdt, #function
- .globl shmdt
- .align 4
- .fnstart
-
-shmdt:
- .save {r4, r7}
- stmfd sp!, {r4, r7}
- ldr r7, =__NR_shmdt
- swi #0
- ldmfd sp!, {r4, r7}
- movs r0, r0
- bxpl lr
- b __set_syscall_errno
- .fnend
diff --git a/libc/arch-arm/syscalls/shmget.S b/libc/arch-arm/syscalls/shmget.S
deleted file mode 100644
index be30704..0000000
--- a/libc/arch-arm/syscalls/shmget.S
+++ /dev/null
@@ -1,19 +0,0 @@
-/* autogenerated by gensyscalls.py */
-#include <sys/linux-syscalls.h>
-
- .text
- .type shmget, #function
- .globl shmget
- .align 4
- .fnstart
-
-shmget:
- .save {r4, r7}
- stmfd sp!, {r4, r7}
- ldr r7, =__NR_shmget
- swi #0
- ldmfd sp!, {r4, r7}
- movs r0, r0
- bxpl lr
- b __set_syscall_errno
- .fnend
diff --git a/libc/bionic/malloc_leak.c b/libc/bionic/malloc_leak.c
index 5ddc913..a0aa2ae 100644
--- a/libc/bionic/malloc_leak.c
+++ b/libc/bionic/malloc_leak.c
@@ -58,6 +58,8 @@
#define SIZE_FLAG_ZYGOTE_CHILD (1<<31)
#define SIZE_FLAG_MASK (SIZE_FLAG_ZYGOTE_CHILD)
+#define MAX_SIZE_T (~(size_t)0)
+
/*
* In a VM process, this is set to 1 after fork()ing out of zygote.
*/
@@ -608,8 +610,16 @@
void* chk_calloc(size_t n_elements, size_t elem_size)
{
- size_t size = n_elements * elem_size;
- void* ptr = chk_malloc(size);
+ size_t size;
+ void* ptr;
+
+ /* Fail on overflow - just to be safe even though this code runs only
+ * within the debugging C library, not the production one */
+ if (n_elements && MAX_SIZE_T / n_elements < elem_size) {
+ return NULL;
+ }
+ size = n_elements * elem_size;
+ ptr = chk_malloc(size);
if (ptr != NULL) {
memset(ptr, 0, size);
}
@@ -763,8 +773,16 @@
void* leak_calloc(size_t n_elements, size_t elem_size)
{
- size_t size = n_elements * elem_size;
- void* ptr = leak_malloc(size);
+ size_t size;
+ void* ptr;
+
+ /* Fail on overflow - just to be safe even though this code runs only
+ * within the debugging C library, not the production one */
+ if (n_elements && MAX_SIZE_T / n_elements < elem_size) {
+ return NULL;
+ }
+ size = n_elements * elem_size;
+ ptr = leak_malloc(size);
if (ptr != NULL) {
memset(ptr, 0, size);
}
diff --git a/libc/bionic/pthread-timers.c b/libc/bionic/pthread-timers.c
index b8f7488..818b47d 100644
--- a/libc/bionic/pthread-timers.c
+++ b/libc/bionic/pthread-timers.c
@@ -469,7 +469,7 @@
}
if ( __likely(!TIMER_ID_IS_WRAPPED(id)) ) {
- return __timer_gettime( id, ospec );
+ return __timer_settime( id, flags, spec, ospec );
} else {
thr_timer_t* timer = thr_timer_from_id(id);
struct timespec expires, now;
@@ -560,11 +560,11 @@
if (timespec_cmp( &expires, &now ) > 0)
{
/* cool, there was no overrun, so compute the
- * relative timeout as 'now - expires', then wait
+ * relative timeout as 'expires - now', then wait
*/
int ret;
- struct timespec diff = now;
- timespec_sub( &diff, &expires );
+ struct timespec diff = expires;
+ timespec_sub( &diff, &now );
ret = __pthread_cond_timedwait_relative(
&timer->cond, &timer->mutex, &diff);
diff --git a/libc/docs/OVERVIEW.TXT b/libc/docs/OVERVIEW.TXT
index 4d40df6..4c153b1 100644
--- a/libc/docs/OVERVIEW.TXT
+++ b/libc/docs/OVERVIEW.TXT
@@ -7,89 +7,93 @@
The core idea behind Bionic's design is: KEEP IT REALLY SIMPLE.
- This implies that the C library should only provide lightweight wrappers around kernel
- facilities and not try to be too smart to deal with edge cases.
+ This implies that the C library should only provide lightweight wrappers
+ around kernel facilities and not try to be too smart to deal with edge cases.
- The name "Bionic" comes from the fact that it is part-BSD and part-Linux: its source
- code consists in a mix of BSD C library pieces with custom Linux-specific bits used
- to deal with threads, processes, signals and a few others things.
+ The name "Bionic" comes from the fact that it is part-BSD and part-Linux:
+ its source code consists in a mix of BSD C library pieces with custom
+ Linux-specific bits used to deal with threads, processes, signals and a few
+ others things.
- All original BSD pieces carry the BSD copyright disclaimer. Bionic-specific bits
- carry the Android Open Source Project copyright disclaimer. And everything is released
- under the BSD license.
+ All original BSD pieces carry the BSD copyright disclaimer. Bionic-specific
+ bits carry the Android Open Source Project copyright disclaimer. And
+ everything is released under the BSD license.
Architectures:
- Bionic currently supports the ARM and x86 instruction sets. In theory, it should be
- possible to support more, but this may require a little work (e.g. adding system
- call IDs to SYSCALLS.TXT, described below, or modifying the dynamic linker).
+ Bionic currently supports the ARM and x86 instruction sets. In theory, it
+ should be possible to support more, but this may require a little work (e.g.
+ adding system call IDs to SYSCALLS.TXT, described below, or modifying the
+ dynamic linker).
- The ARM-specific code is under arch-arm/ and the x86-specific one is under arch-x86/
+ The ARM-specific code is under arch-arm/ and the x86-specific one is under
+ arch-x86/
- Note that the x86 version is only meant to run on an x86 Android device. We make
- absolutely no claim that you could build and use Bionic on a stock x86 Linux
- distribution (though that would be cool, so patches are welcomed :-))
+ Note that the x86 version is only meant to run on an x86 Android device. We
+ make absolutely no claim that you could build and use Bionic on a stock x86
+ Linux distribution (though that would be cool, so patches are welcomed :-))
Syscall stubs:
Each system call function is implemented by a tiny assembler source fragment
- (called a "syscall stub"), which is generated automatically by tools/gensyscalls.py
- which reads the SYSCALLS.TXT file for input.
+ (called a "syscall stub"), which is generated automatically by
+ tools/gensyscalls.py which reads the SYSCALLS.TXT file for input.
SYSCALLS.TXT contains the list of all syscall stubs to generate, along with
- the corresponding syscall numeric identifier (which may differ between ARM and x86),
- and its signature
+ the corresponding syscall numeric identifier (which may differ between ARM
+ and x86), and its signature
- If you modify this file, you may want to use tools/checksyscalls.py which checks
- its content against official Linux kernel header files, and will report errors when
- invalid syscall ids are used.
+ If you modify this file, you may want to use tools/checksyscalls.py which
+ checks its content against official Linux kernel header files, and will
+ report errors when invalid syscall ids are used.
- Sometimes, the C library function is really a wrapper that calls the corresponding
- syscall with another name. For example, the exit() function is provided by the C
- library and calls the _exit() syscall stub.
+ Sometimes, the C library function is really a wrapper that calls the
+ corresponding syscall with another name. For example, the exit() function
+ is provided by the C library and calls the _exit() syscall stub.
See SYSCALLS.TXT for documentation and details.
time_t:
- time_t is 32-bit as defined by the kernel on 32-bit CPUs. A 64-bit version would
- be preferrable to avoid the Y2038 bug, but the kernel maintainers consider that
- this is not needed at the moment.
+ time_t is 32-bit as defined by the kernel on 32-bit CPUs. A 64-bit version
+ would be preferrable to avoid the Y2038 bug, but the kernel maintainers
+ consider that this is not needed at the moment.
- Instead, Bionic provides a <time64.h> header that defines a time64_t type, and
- related functions like mktime64(), localtime64(), etc...
+ Instead, Bionic provides a <time64.h> header that defines a time64_t type,
+ and related functions like mktime64(), localtime64(), etc...
Timezone management:
- The name of the current timezone is taken from the TZ environment variable, if defined.
- Otherwise, the system property named 'persist.sys.timezone' is checked instead.
+ The name of the current timezone is taken from the TZ environment variable,
+ if defined. Otherwise, the system property named 'persist.sys.timezone' is
+ checked instead.
The zoneinfo timezone database and index files are located under directory
- /system/usr/share/zoneinfo, instead of the more Posix path of /usr/share/zoneinfo
+ /system/usr/share/zoneinfo, instead of the more Posix-compliant path of
+ /usr/share/zoneinfo
off_t:
- For similar reasons, off_t is 32-bit. We define loff_t as the 64-bit variant due
- to BSD inheritance, but off64_t should be available as a typedef to ease porting of
- current Linux-specific code.
-
+ For similar reasons, off_t is 32-bit. We define loff_t as the 64-bit variant
+ due to BSD inheritance, but off64_t should be available as a typedef to ease
+ porting of current Linux-specific code.
Linux kernel headers:
- Bionic comes with its own set of "clean" Linux kernel headers to allow user-space
- code to use kernel-specific declarations (e.g. IOCTLs, structure declarations,
- constants, etc...). They are located in:
+ Bionic comes with its own set of "clean" Linux kernel headers to allow
+ user-space code to use kernel-specific declarations (e.g. IOCTLs, structure
+ declarations, constants, etc...). They are located in:
./kernel/common,
./kernel/arch-arm
./kernel/arch-x86
- These headers have been generated by a tool (kernel/tools/update-all.py) to only
- include the public definitions from the original Linux kernel headers.
+ These headers have been generated by a tool (kernel/tools/update-all.py) to
+ only include the public definitions from the original Linux kernel headers.
If you want to know why and how this is done, read kernel/README.TXT to get
all the (gory) details.
@@ -97,21 +101,23 @@
PThread implementation:
- Bionic's C library comes with its own pthread implementation bundled in. This is
- different from other historical C libraries which:
+ Bionic's C library comes with its own pthread implementation bundled in.
+ This is different from other historical C libraries which:
- place it in an external library (-lpthread)
- play linker tricks with weak symbols at dynamic link time
- The support for real-time features (a.k.a. -lrt) is also bundled in the C library.
+ The support for real-time features (a.k.a. -lrt) is also bundled in the
+ C library.
- The implementation is based on futexes and strives to provide *very* short code paths
- for common operations. Notable features are the following:
+ The implementation is based on futexes and strives to provide *very* short
+ code paths for common operations. Notable features are the following:
- pthread_mutex_t, pthread_cond_t are only 4 bytes each.
- - Normal, recursive and error-check mutexes are supported, and the code path
- is heavily optimized for the normal case, which is used most of the time.
+ - Normal, recursive and error-check mutexes are supported, and the code
+ path is heavily optimized for the normal case, which is used most of
+ the time.
- Process-shared mutexes and condition variables are not supported.
Their implementation requires far more complexity and was absolutely
@@ -123,8 +129,9 @@
paths slightly slower though).
- There is currently no support for read/write locks, priority-ceiling in
- mutexes and other more advanced features. Again, the main idea being that
- this was not needed for Android at all but could be added in the future.
+ mutexes and other more advanced features. Again, the main idea being
+ that this was not needed for Android at all but could be added in the
+ future.
pthread_cancel():
@@ -134,13 +141,13 @@
Consider that:
- A proper implementation must insert pthread cancellation checks in a lot
- of different places of the C library. And conformance is very difficult to
- test properly.
+ of different places of the C library. And conformance is very difficult
+ to test properly.
- - A proper implementation must also clean up resources, like releasing memory,
- or unlocking mutexes, properly if the cancellation happens in a complex
- function (e.g. inside gethostbyname() or fprintf() + complex formatting
- rules). This tends to slow down the path of many functions.
+ - A proper implementation must also clean up resources, like releasing
+ memory, or unlocking mutexes, properly if the cancellation happens in a
+ complex function (e.g. inside gethostbyname() or fprintf() + complex
+ formatting rules). This tends to slow down the path of many functions.
- pthread cancellation cannot stop all threads: e.g. it can't do anything
against an infinite loop
@@ -151,18 +158,20 @@
All of this is contrary to the Bionic design goals. If your code depends on
thread cancellation, please consider alternatives.
- Note however that Bionic does implement pthread_cleanup_push() and pthread_cleanup_pop(),
- which can be used to handle cleanups that happen when a thread voluntarily exits
- through pthread_exit() or returning from its main function.
+ Note however that Bionic does implement pthread_cleanup_push() and
+ pthread_cleanup_pop(), which can be used to handle cleanups that happen when
+ a thread voluntarily exits through pthread_exit() or returning from its
+ main function.
pthread_once():
Do not call fork() within a callback provided to pthread_once(). Doing this
- may result in a deadlock in the child process the next time it calls pthread_once().
+ may result in a deadlock in the child process the next time it calls
+ pthread_once().
- Also, you can't throw a C++ Exception from the callback (see C++ Exception Support
- below).
+ Also, you can't throw a C++ Exception from the callback (see C++ Exception
+ Support below).
The current implementation of pthread_once() lacks the necessary support of
multi-core-safe double-checked-locking (read and write barriers).
@@ -170,27 +179,27 @@
Thread-specific data
- The thread-specific storage only provides for a bit less than 64 pthread_key_t
- objects to each process. The implementation provides 64 real slots but also
- uses about 5 of them (exact number may depend on implementation) for its
- own use (e.g. two slots are pre-allocated by the C library to speed-up the
- Android OpenGL sub-system).
+ The thread-specific storage only provides for a bit less than 64
+ pthread_key_t objects to each process. The implementation provides 64 real
+ slots but also uses about 5 of them (exact number may depend on
+ implementation) for its own use (e.g. two slots are pre-allocated by the C
+ library to speed-up the Android OpenGL sub-system).
Note that Posix mandates a minimum of 128 slots, but we do not claim to be
Posix-compliant.
- Except for the main thread, the TLS area is stored at the top of the stack. See
- comments in bionic/libc/bionic/pthread.c for details.
+ Except for the main thread, the TLS area is stored at the top of the stack.
+ See comments in bionic/libc/bionic/pthread.c for details.
- At the moment, thread-local storage defined through the __thread compiler keyword
- is not supported by the Bionic C library and dynamic linker.
+ At the moment, thread-local storage defined through the __thread compiler
+ keyword is not supported by the Bionic C library and dynamic linker.
Multi-core support
At the moment, Bionic does not provide or use read/write memory barriers.
- This means that using it on certain multi-core systems might not be supported,
- depending on its exact CPU architecture.
+ This means that using it on certain multi-core systems might not be
+ supported, depending on its exact CPU architecture.
Android-specific features:
@@ -201,67 +210,69 @@
Android provides a simple shared value/key space to all processes on the
system. It stores a liberal number of 'properties', each of them being a
- simple size-limited string that can be associated to a size-limited string
- value.
+ simple size-limited string that can be associated to a size-limited
+ string value.
- The header <sys/system_properties.h> can be used to read system properties
- and also defines the maximum size of keys and values.
+ The header <sys/system_properties.h> can be used to read system
+ properties and also defines the maximum size of keys and values.
- Android-specific user/group management:
- There is no /etc/passwd or /etc/groups in Android. By design, it is meant to
- be used by a single handset user. On the other hand, Android uses the Linux
- user/group management features extensively to secure process permissions,
- like access to various filesystem directories.
+ There is no /etc/passwd or /etc/groups in Android. By design, it is
+ meant to be used by a single handset user. On the other hand, Android
+ uses the Linux user/group management features extensively to secure
+ process permissions, like access to various filesystem directories.
- In the Android scheme, each installed application gets its own uid_t/gid_t
- starting from 10000; lower numerical ids are reserved for system daemons.
+ In the Android scheme, each installed application gets its own
+ uid_t/gid_t starting from 10000; lower numerical ids are reserved for
+ system daemons.
- getpwnam() recognizes some hard-coded subsystems names (e.g. "radio") and
- will translate them to their low-user-id values. It also recognizes "app_1234"
- as the synthetic name of the application that was installed with uid 10000 + 1234,
- which is 11234. getgrnam() works similarly
+ getpwnam() recognizes some hard-coded subsystems names (e.g. "radio")
+ and will translate them to their low-user-id values. It also recognizes
+ "app_1234" as the synthetic name of the application that was installed
+ with uid 10000 + 1234, which is 11234. getgrnam() works similarly
- getgrouplist() will always return a single group for any user name, which is
- the one passed as an input parameter.
+ getgrouplist() will always return a single group for any user name,
+ which is the one passed as an input parameter.
- getgrgid() will similarly only return a structure that contains a single-element
- members list, corresponding to the user with the same numerical value than the
- group.
+ getgrgid() will similarly only return a structure that contains a
+ single-element members list, corresponding to the user with the same
+ numerical value than the group.
See bionic/libc/bionic/stubs.c for more details.
- getservent()
- There is no /etc/services on Android. Instead the C library embeds a constant
- list of services in its executable, which is parsed on demand by the various
- functions that depend on it. See bionic/libc/netbsd/net/getservent.c and
+ There is no /etc/services on Android. Instead the C library embeds a
+ constant list of services in its executable, which is parsed on demand
+ by the various functions that depend on it. See
+ bionic/libc/netbsd/net/getservent.c and
bionic/libc/netbsd/net/services.h
- The list of services defined internally might change liberally in the future.
- This feature is mostly historically and is very rarely used.
+ The list of services defined internally might change liberally in the
+ future. This feature is mostly historically and is very rarely used.
- The getservent() returns thread-local data. getservbyport() and getservbyname()
- are also implemented in a similar fashion.
+ The getservent() returns thread-local data. getservbyport() and
+ getservbyname() are also implemented in a similar fashion.
- getprotoent()
- There is no /etc/protocol on Android. Bionic does not currently implement
- getprotoent() and related functions. If we add it, it will likely be done
- in a way similar to getservent()
+ There is no /etc/protocol on Android. Bionic does not currently
+ implement getprotoent() and related functions. If added, it will
+ likely be done in a way similar to getservent()
DNS resolver:
- Bionic uses a NetBSD-derived resolver library which has been modified in the following
- ways:
+ Bionic uses a NetBSD-derived resolver library which has been modified in
+ the following ways:
- don't implement the name-server-switch feature (a.k.a. <nsswitch.h>)
- read /system/etc/resolv.conf instead of /etc/resolv.conf
- read the list of servers from system properties. the code looks for
- 'net.dns1', 'net.dns2', etc.. Each property should contain the IP address
- of a DNS server.
+ 'net.dns1', 'net.dns2', etc.. Each property should contain the IP
+ address of a DNS server.
these properties are set/modified by other parts of the Android system
(e.g. the dhcpd daemon).
@@ -278,9 +289,9 @@
- get rid of *many* unfortunate thread-safety issues in the original code
- Bionic does *not* expose implementation details of its DNS resolver; the content
- of <arpa/nameser.h> is intentionally blank. The resolver implementation might
- change completely in the future.
+ Bionic does *not* expose implementation details of its DNS resolver; the
+ content of <arpa/nameser.h> is intentionally blank. The resolver
+ implementation might change completely in the future.
PThread Real-Time Timers:
@@ -294,8 +305,8 @@
timers with compatible properties are used.
This means that if your code uses a lot of SIGEV_THREAD timers, your program
- may consume a lot of memory. However, if your program needs many of these timers,
- it'd better handle timeout events directly instead.
+ may consume a lot of memory. However, if your program needs many of these
+ timers, it'd better handle timeout events directly instead.
Other timers (e.g. SIGEV_SIGNAL) are handled by the kernel and use much less
system resources.
@@ -303,58 +314,71 @@
Binary Compatibility:
- Bionic is *not* in any way binary-compatible with the GNU C Library, ucLibc or any
- known Linux C library. This means several things:
+ Bionic is *not* in any way binary-compatible with the GNU C Library, ucLibc
+ or any known Linux C library. This means several things:
- - You cannot expect to build something against the GNU C Library headers and have
- it dynamically link properly to Bionic later.
+ - You cannot expect to build something against the GNU C Library headers and
+ have it dynamically link properly to Bionic later.
- - You should *really* use the Android toolchain to build your program against Bionic.
- The toolchain deals with many important details that are crucial to get something
- working properly.
+ - You should *really* use the Android toolchain to build your program against
+ Bionic. The toolchain deals with many important details that are crucial
+ to get something working properly.
- Failure to do so will usually result in the inability to run or link your program,
- or even runtime crashes. Several random web pages on the Internet describe how you
- can succesfully write a "hello-world" program with the ARM GNU toolchain. These
- examples usually work by chance, if anything else, and you should not follow these
- instructions unless you want to waste a lot of your time in the process.
+ Failure to do so will usually result in the inability to run or link your
+ program, or even runtime crashes. Several random web pages on the Internet
+ describe how you can succesfully write a "hello-world" program with the
+ ARM GNU toolchain. These examples usually work by chance, if anything else,
+ and you should not follow these instructions unless you want to waste a lot
+ of your time in the process.
- Note however that you *can* generate a binary that is built against the GNU C Library
- headers and then statically linked to it. The corresponding executable should be able
- to run (if it doesn't use dlopen()/dlsym())
+ Note however that you *can* generate a binary that is built against the
+ GNU C Library headers and then statically linked to it. The corresponding
+ executable should be able to run (if it doesn't use dlopen()/dlsym())
+
Dynamic Linker:
- Bionic comes with its own dynamic linker (just like ld.so on Linux really comes from
- GLibc). This linker does not support all the relocations generated by other GCC ARM
- toolchains.
+ Bionic comes with its own dynamic linker (just like ld.so on Linux really
+ comes from GLibc). This linker does not support all the relocations
+ generated by other GCC ARM toolchains.
+
C++ Exceptions Support:
- At the moment, Bionic doesn't support C++ exceptions, what this really means is the
- following:
+ At the moment, Bionic doesn't support C++ exceptions, what this really means
+ is the following:
- If pthread_once() is called with a C++ callback that throws an exception,
- then the C library will keep the corresponding pthread_once_t mutex locked.
- Any further call to pthread_once() will result in a deadlock.
+ then the C library will keep the corresponding pthread_once_t mutex
+ locked. Any further call to pthread_once() will result in a deadlock.
- A proper implementation should be able to register a C++ exception cleanup
- handler before the callback to properly unlock the pthread_once_t. Unfortunately
- this requires tricky assembly code that is highly dependent on the compiler.
+ A proper implementation should be able to register a C++ exception
+ cleanup handler before the callback to properly unlock the
+ pthread_once_t. Unfortunately this requires tricky assembly code that
+ is highly dependent on the compiler.
This feature is not planned to be supported anytime soon.
- - The same problem may arise if you throw an exception within a callback called
- from the C library. Fortunately, these cases are very rare in the real-world,
- but any callback you provide to the C library should *not* throw an exception.
+ - The same problem may arise if you throw an exception within a callback
+ called from the C library. Fortunately, these cases are very rare in the
+ real-world, but any callback you provide to the C library should *not*
+ throw an exception.
- - Bionic lacks a few support functions to have exception support work properly.
+ - Bionic lacks a few support functions to have exception support work
+ properly.
+
+System V IPCs:
+
+ Bionic intentionally does not provide support for System-V IPCs mechanisms,
+ like the ones provided by semget(), shmget(), msgget(). The reason for this
+ is to avoid denial-of-service. For a detailed rationale about this, please
+ read the file docs/SYSV-IPCS.TXT.
Include Paths:
- The Android build system should automatically provide the necessary include paths
- required to build against the C library headers. However, if you want to do that
- yourself, you will need to add:
+ The Android build system should automatically provide the necessary include
+ paths required to build against the C library headers. However, if you want
+ to do that yourself, you will need to add:
libc/arch-$ARCH/include
libc/include
diff --git a/libc/docs/SYSV-IPC.TXT b/libc/docs/SYSV-IPC.TXT
new file mode 100644
index 0000000..5a3eef0
--- /dev/null
+++ b/libc/docs/SYSV-IPC.TXT
@@ -0,0 +1,103 @@
+Android does not support System V IPCs, i.e. the facilities provided by the
+following standard Posix headers:
+
+ <sys/sem.h> /* SysV semaphores */
+ <sys/shm.h> /* SysV shared memory segments */
+ <sys/msg.h> /* SysV message queues */
+ <sys/ipc.h> /* General IPC definitions */
+
+The reason for this is due to the fact that, by design, they lead to global
+kernel resource leakage.
+
+For example, there is no way to automatically release a SysV semaphore
+allocated in the kernel when:
+
+- a buggy or malicious process exits
+- a non-buggy and non-malicious process crashes or is explicitely killed.
+
+Killing processes automatically to make room for new ones is an
+important part of Android's application lifecycle implementation. This means
+that, even assuming only non-buggy and non-malicious code, it is very likely
+that over time, the kernel global tables used to implement SysV IPCs will fill
+up.
+
+At that point, strange failures are likely to occur and prevent programs that
+use them to run properly until the next reboot of the system.
+
+And we can't ignore potential malicious applications. As a proof of concept
+here is a simple exploit that you can run on a standard Linux box today:
+
+--------------- cut here ------------------------
+#include <sys/sem.h>
+#include <sys/wait.h>
+#include <unistd.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <errno.h>
+
+#define NUM_SEMAPHORES 32
+#define MAX_FAILS 10
+
+int main(void)
+{
+ int counter = 0;
+ int fails = 0;
+
+ if (counter == IPC_PRIVATE)
+ counter++;
+
+ printf( "%d (NUM_SEMAPHORES=%d)\n", counter, NUM_SEMAPHORES);
+
+ for (;;) {
+ int ret = fork();
+ int status;
+
+ if (ret < 0) {
+ perror("fork:");
+ break;
+ }
+ if (ret == 0) {
+ /* in the child */
+ ret = semget( (key_t)counter, NUM_SEMAPHORES, IPC_CREAT );
+ if (ret < 0) {
+ return errno;
+ }
+ return 0;
+ }
+ else {
+ /* in the parent */
+ ret = wait(&status);
+ if (ret < 0) {
+ perror("waitpid:");
+ break;
+ }
+ if (status != 0) {
+ status = WEXITSTATUS(status);
+ fprintf(stderr, "child %d FAIL at counter=%d: %d\n", ret,
+ counter, status);
+ if (++fails >= MAX_FAILS)
+ break;
+ }
+ }
+
+ counter++;
+ if ((counter % 1000) == 0) {
+ printf("%d\n", counter);
+ }
+ if (counter == IPC_PRIVATE)
+ counter++;
+ }
+ return 0;
+}
+--------------- cut here ------------------------
+
+If you run it on a typical Linux distribution today, you'll discover that it
+will quickly fill up the kernel's table of unique key_t values, and that
+strange things will happen in some parts of the system, but not all.
+
+(You can use the "ipcs -u" command to get a summary describing the kernel
+ tables and their allocations)
+
+For example, in our experience, anything program launched after that that
+calls strerror() will simply crash. The USB sub-system starts spoutting weird
+errors to the system console, etc...
diff --git a/libc/include/sys/linux-syscalls.h b/libc/include/sys/linux-syscalls.h
index bed00ce..7772f1e 100644
--- a/libc/include/sys/linux-syscalls.h
+++ b/libc/include/sys/linux-syscalls.h
@@ -170,17 +170,6 @@
#define __NR_getsockopt (__NR_SYSCALL_BASE + 295)
#define __NR_sendmsg (__NR_SYSCALL_BASE + 296)
#define __NR_recvmsg (__NR_SYSCALL_BASE + 297)
-#define __NR_semctl (__NR_SYSCALL_BASE + 300)
-#define __NR_semget (__NR_SYSCALL_BASE + 299)
-#define __NR_semop (__NR_SYSCALL_BASE + 298)
-#define __NR_shmat (__NR_SYSCALL_BASE + 305)
-#define __NR_shmctl (__NR_SYSCALL_BASE + 308)
-#define __NR_shmdt (__NR_SYSCALL_BASE + 306)
-#define __NR_shmget (__NR_SYSCALL_BASE + 307)
-#define __NR_msgctl (__NR_SYSCALL_BASE + 304)
-#define __NR_msgget (__NR_SYSCALL_BASE + 303)
-#define __NR_msgrcv (__NR_SYSCALL_BASE + 302)
-#define __NR_msgsnd (__NR_SYSCALL_BASE + 301)
#define __NR_epoll_create (__NR_SYSCALL_BASE + 250)
#define __NR_epoll_ctl (__NR_SYSCALL_BASE + 251)
#define __NR_epoll_wait (__NR_SYSCALL_BASE + 252)
diff --git a/libc/include/sys/linux-unistd.h b/libc/include/sys/linux-unistd.h
index 8539c7b..789271e 100644
--- a/libc/include/sys/linux-unistd.h
+++ b/libc/include/sys/linux-unistd.h
@@ -179,17 +179,6 @@
int sched_get_priority_max (int policy);
int sched_get_priority_min (int policy);
int sched_rr_get_interval (pid_t pid, struct timespec *interval);
-int semctl (int semid, int semnum, int cmd, ...);
-int semget (key_t key, int nsems, int semflg);
-int semop (int semid, struct sembuf* sops, size_t nsops);
-void* shmat (int shmid, const void* shmaddr, int shmflg);
-int shmctl (int shmid, int cmd, struct shmid_ds* buf);
-int shmdt (const void* shmaddr);
-int shmget (key_t key, size_t size, int shmflg);
-int msgctl (int msqid, int cmd, struct msqid_ds *buf);
-int msgget (key_t key, int msgflg);
-int msgrcv (int msqid, void* msgp, size_t msgsz, long int msgtyp, int msgflg);
-int msgsnd (int msqid, const void* msgp, size_t msgsz, int msgflg);
int uname (struct utsname *);
pid_t __wait4 (pid_t pid, int *status, int options, struct rusage *rusage);
mode_t umask (mode_t);
diff --git a/libc/include/sys/msg.h b/libc/include/sys/msg.h
deleted file mode 100644
index dd1b527..0000000
--- a/libc/include/sys/msg.h
+++ /dev/null
@@ -1,43 +0,0 @@
-/*
- * Copyright (C) 2008 The Android Open Source Project
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- * * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- * * Redistributions in binary form must reproduce the above copyright
- * notice, this list of conditions and the following disclaimer in
- * the documentation and/or other materials provided with the
- * distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
- * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
- * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
- * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
- * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS
- * OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
- * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
- * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
- * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
- * SUCH DAMAGE.
- */
-#ifndef _SYS_MSG_H
-#define _SYS_MSG_H
-
-#include <sys/ipc.h>
-#include <linux/msg.h>
-
-__BEGIN_DECLS
-
-extern int msgctl(int msqid, int cmd, struct msqid_ds *buf);
-extern int msgget(key_t key, int msgflg);
-extern int msgrcv(int msqid, void* msgp, size_t msgsz, long int msgtyp, int msgflg);
-extern int msgsnd(int msqid, const void* msgp, size_t msgsz, int msgflg);
-
-__END_DECLS
-
-#endif /* _SYS_MSG_H */
diff --git a/libc/include/sys/shm.h b/libc/include/sys/shm.h
deleted file mode 100644
index 495d33e..0000000
--- a/libc/include/sys/shm.h
+++ /dev/null
@@ -1,43 +0,0 @@
-/*
- * Copyright (C) 2008 The Android Open Source Project
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- * * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- * * Redistributions in binary form must reproduce the above copyright
- * notice, this list of conditions and the following disclaimer in
- * the documentation and/or other materials provided with the
- * distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
- * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
- * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
- * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
- * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS
- * OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
- * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
- * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
- * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
- * SUCH DAMAGE.
- */
-#ifndef _SYS_SHM_H
-#define _SYS_SHM_H
-
-#include <sys/ipc.h>
-#include <linux/shm.h>
-
-__BEGIN_DECLS
-
-extern void* shmat(int shmid, const void* shmaddr, int shmflg);
-extern int shmctl(int shmid, int cmd, struct shmid_ds* buf);
-extern int shmdt(const void* shmaddr);
-extern int shmget(key_t key, size_t size, int shmflg);
-
-__END_DECLS
-
-#endif /* _SYS_SHM_H */
diff --git a/libc/kernel/common/linux/msm_audio.h b/libc/kernel/common/linux/msm_audio.h
index 58dc6f3..9ac58aa 100644
--- a/libc/kernel/common/linux/msm_audio.h
+++ b/libc/kernel/common/linux/msm_audio.h
@@ -29,6 +29,10 @@
#define AUDIO_SET_EQ _IOW(AUDIO_IOCTL_MAGIC, 8, unsigned)
#define AUDIO_SET_RX_IIR _IOW(AUDIO_IOCTL_MAGIC, 9, unsigned)
#define AUDIO_SET_VOLUME _IOW(AUDIO_IOCTL_MAGIC, 10, unsigned)
+#define AUDIO_ENABLE_AUDPRE _IOW(AUDIO_IOCTL_MAGIC, 11, unsigned)
+#define AUDIO_SET_AGC _IOW(AUDIO_IOCTL_MAGIC, 12, unsigned)
+#define AUDIO_SET_NS _IOW(AUDIO_IOCTL_MAGIC, 13, unsigned)
+#define AUDIO_SET_TX_IIR _IOW(AUDIO_IOCTL_MAGIC, 14, unsigned)
struct msm_audio_config {
uint32_t buffer_size;
diff --git a/libc/netbsd/gethnamaddr.c b/libc/netbsd/gethnamaddr.c
index e6f919e..1c219b2 100644
--- a/libc/netbsd/gethnamaddr.c
+++ b/libc/netbsd/gethnamaddr.c
@@ -626,37 +626,12 @@
break;
}
-#ifdef ANDROID_CHANGES
- cache = __get_res_cache();
- if (cache != NULL) {
- hp = _resolv_cache_lookup( cache, name, af );
- if (hp == _RESOLV_HOSTENT_NONE) {
- h_errno = HOST_NOT_FOUND;
- return NULL;
- }
- if (hp != NULL) {
- h_errno = NETDB_SUCCESS;
- return hp;
- }
- }
-#endif
-
hp = NULL;
h_errno = NETDB_INTERNAL;
if (nsdispatch(&hp, dtab, NSDB_HOSTS, "gethostbyname",
default_dns_files, name, strlen(name), af) != NS_SUCCESS) {
-#ifdef ANDROID_CHANGES
- /* cache negative DNS entry */
- if (h_errno == HOST_NOT_FOUND && cache != NULL)
- _resolv_cache_add( cache, name, af, _RESOLV_HOSTENT_NONE );
-#endif
return NULL;
}
-#ifdef ANDROID_CHANGES
- if (cache != NULL) {
- _resolv_cache_add( cache, name, af, hp );
- }
-#endif
h_errno = NETDB_SUCCESS;
return hp;
}
diff --git a/libc/netbsd/resolv/res_cache.c b/libc/netbsd/resolv/res_cache.c
index 210dfdd..2c912de 100644
--- a/libc/netbsd/resolv/res_cache.c
+++ b/libc/netbsd/resolv/res_cache.c
@@ -28,30 +28,104 @@
#include "resolv_cache.h"
#include <stdlib.h>
+#include <string.h>
#include <time.h>
#include "pthread.h"
-/* this code implements a small DNS resolver cache for all gethostbyname* functions
+/* This code implements a small and *simple* DNS resolver cache.
*
- * the cache is shared between all threads of the current process, but the result of
- * a succesful lookup is always copied to a thread-local variable to honor the persistence
- * rules of the gethostbyname*() APIs.
+ * It is only used to cache DNS answers for a maximum of CONFIG_SECONDS seconds
+ * in order to reduce DNS traffic. It is not supposed to be a full DNS cache,
+ * since we plan to implement that in the future in a dedicated process running
+ * on the system.
+ *
+ * Note that its design is kept simple very intentionally, i.e.:
+ *
+ * - it takes raw DNS query packet data as input, and returns raw DNS
+ * answer packet data as output
+ *
+ * (this means that two similar queries that encode the DNS name
+ * differently will be treated distinctly).
+ *
+ * - the TTLs of answer RRs are ignored. our DNS resolver library does not use
+ * them anyway, but it means that records with a TTL smaller than
+ * CONFIG_SECONDS will be kept in the cache anyway.
+ *
+ * this is bad, but we absolutely want to avoid parsing the answer packets
+ * (and should be solved by the later full DNS cache process).
+ *
+ * - the implementation is just a (query-data) => (answer-data) hash table
+ * with a trivial least-recently-used expiration policy.
+ *
+ * Doing this keeps the code simple and avoids to deal with a lot of things
+ * that a full DNS cache is expected to do.
+ *
+ * The API is also very simple:
+ *
+ * - the client calls _resolv_cache_get() to obtain a handle to the cache.
+ * this will initialize the cache on first usage. the result can be NULL
+ * if the cache is disabled.
+ *
+ * - the client calls _resolv_cache_lookup() before performing a query
+ *
+ * if the function returns RESOLV_CACHE_FOUND, a copy of the answer data
+ * has been copied into the client-provided answer buffer.
+ *
+ * if the function returns RESOLV_CACHE_NOTFOUND, the client should perform
+ * a request normally, *then* call _resolv_cache_add() to add the received
+ * answer to the cache.
+ *
+ * if the function returns RESOLV_CACHE_UNSUPPORTED, the client should
+ * perform a request normally, and *not* call _resolv_cache_add()
+ *
+ * note that RESOLV_CACHE_UNSUPPORTED is also returned if the answer buffer
+ * is too short to accomodate the cached result.
+ *
+ * - when network settings change, the cache must be flushed since the list
+ * of DNS servers probably changed. this is done by calling
+ * _resolv_cache_reset()
+ *
+ * the parameter to this function must be an ever-increasing generation
+ * number corresponding to the current network settings state.
+ *
+ * This is done because several threads could detect the same network
+ * settings change (but at different times) and will all end up calling the
+ * same function. Comparing with the last used generation number ensures
+ * that the cache is only flushed once per network change.
*/
-/* the name of an environment variable that will be checked the first time this code is called
- * if its value is "0", then the resolver cache is disabled.
+/* the name of an environment variable that will be checked the first time
+ * this code is called if its value is "0", then the resolver cache is
+ * disabled.
*/
#define CONFIG_ENV "BIONIC_DNSCACHE"
-/* entries older than CONFIG_SECONDS seconds are always discarded. otherwise we could keep
- * stale addresses in the cache and be oblivious to DNS server changes
+/* entries older than CONFIG_SECONDS seconds are always discarded.
*/
#define CONFIG_SECONDS (60*10) /* 10 minutes */
-/* maximum number of entries kept in the cache. be frugal, this is the C library
- * this MUST BE A POWER OF 2
+/* maximum number of entries kept in the cache. This value has been
+ * determined by browsing through various sites and counting the number
+ * of corresponding requests. Keep in mind that our framework is currently
+ * performing two requests per name lookup (one for IPv4, the other for IPv6)
+ *
+ * www.google.com 4
+ * www.ysearch.com 6
+ * www.amazon.com 8
+ * www.nytimes.com 22
+ * www.espn.com 28
+ * www.msn.com 28
+ * www.lemonde.fr 35
+ *
+ * (determined in 2009-2-17 from Paris, France, results may vary depending
+ * on location)
+ *
+ * most high-level websites use lots of media/ad servers with different names
+ * but these are generally reused when browsing through the site.
+ *
+ * As such, a valud of 64 should be relatively conformtable at the moment.
*/
-#define CONFIG_MAX_ENTRIES 128
+#define CONFIG_MAX_ENTRIES 64
/****************************************************************************/
/****************************************************************************/
@@ -61,42 +135,204 @@
/****************************************************************************/
/****************************************************************************/
-#define DEBUG 0
+/* set to 1 to debug cache operations */
+#define DEBUG 0
+
+/* set to 1 to debug query data */
+#define DEBUG_DATA 0
#if DEBUG
-#include <fcntl.h>
-#include <errno.h>
-#include <stdarg.h>
+# include <logd.h>
+# define XLOG(...) \
+ __libc_android_log_print(ANDROID_LOG_DEBUG,"libc",__VA_ARGS__)
+
#include <stdio.h>
-static void
-xlog( const char* fmt, ... )
+#include <stdarg.h>
+
+/** BOUNDED BUFFER FORMATTING
+ **/
+
+/* technical note:
+ *
+ * the following debugging routines are used to append data to a bounded
+ * buffer they take two parameters that are:
+ *
+ * - p : a pointer to the current cursor position in the buffer
+ * this value is initially set to the buffer's address.
+ *
+ * - end : the address of the buffer's limit, i.e. of the first byte
+ * after the buffer. this address should never be touched.
+ *
+ * IMPORTANT: it is assumed that end > buffer_address, i.e.
+ * that the buffer is at least one byte.
+ *
+ * the _bprint_() functions return the new value of 'p' after the data
+ * has been appended, and also ensure the following:
+ *
+ * - the returned value will never be strictly greater than 'end'
+ *
+ * - a return value equal to 'end' means that truncation occured
+ * (in which case, end[-1] will be set to 0)
+ *
+ * - after returning from a _bprint_() function, the content of the buffer
+ * is always 0-terminated, even in the event of truncation.
+ *
+ * these conventions allow you to call _bprint_ functions multiple times and
+ * only check for truncation at the end of the sequence, as in:
+ *
+ * char buff[1000], *p = buff, *end = p + sizeof(buff);
+ *
+ * p = _bprint_c(p, end, '"');
+ * p = _bprint_s(p, end, my_string);
+ * p = _bprint_c(p, end, '"');
+ *
+ * if (p >= end) {
+ * // buffer was too small
+ * }
+ *
+ * printf( "%s", buff );
+ */
+
+/* add a char to a bounded buffer */
+static char*
+_bprint_c( char* p, char* end, int c )
{
- static int fd = -2;
- int ret;
-
- if (fd == -2) {
- do {
- fd = open( "/data/dns.log", O_CREAT | O_APPEND | O_WRONLY, 0666 );
- } while (fd < 0 && errno == EINTR);
+ if (p < end) {
+ if (p+1 == end)
+ *p++ = 0;
+ else {
+ *p++ = (char) c;
+ *p = 0;
+ }
}
-
- if (fd >= 0) {
- char temp[128];
- va_list args;
- va_start(args, fmt);
- vsnprintf( temp, sizeof(temp), fmt, args);
- va_end(args);
-
- do {
- ret = write( fd, temp, strlen(temp) );
- } while (ret == -1 && errno == EINTR);
- }
+ return p;
}
-#define XLOG(...) xlog(__VA_ARGS__)
-#else
-#define XLOG(...) ((void)0)
-#endif
+/* add a sequence of bytes to a bounded buffer */
+static char*
+_bprint_b( char* p, char* end, const char* buf, int len )
+{
+ int avail = end - p;
+
+ if (avail <= 0 || len <= 0)
+ return p;
+
+ if (avail > len)
+ avail = len;
+
+ memcpy( p, buf, avail );
+ p += avail;
+
+ if (p < end)
+ p[0] = 0;
+ else
+ end[-1] = 0;
+
+ return p;
+}
+
+/* add a string to a bounded buffer */
+static char*
+_bprint_s( char* p, char* end, const char* str )
+{
+ return _bprint_b(p, end, str, strlen(str));
+}
+
+/* add a formatted string to a bounded buffer */
+static char*
+_bprint( char* p, char* end, const char* format, ... )
+{
+ int avail, n;
+ va_list args;
+
+ avail = end - p;
+
+ if (avail <= 0)
+ return p;
+
+ va_start(args, format);
+ n = snprintf( p, avail, format, args);
+ va_end(args);
+
+ /* certain C libraries return -1 in case of truncation */
+ if (n < 0 || n > avail)
+ n = avail;
+
+ p += n;
+ /* certain C libraries do not zero-terminate in case of truncation */
+ if (p == end)
+ p[-1] = 0;
+
+ return p;
+}
+
+/* add a hex value to a bounded buffer, up to 8 digits */
+static char*
+_bprint_hex( char* p, char* end, unsigned value, int numDigits )
+{
+ char text[sizeof(unsigned)*2];
+ int nn = 0;
+
+ while (numDigits-- > 0) {
+ text[nn++] = "0123456789abcdef"[(value >> (numDigits*4)) & 15];
+ }
+ return _bprint_b(p, end, text, nn);
+}
+
+/* add the hexadecimal dump of some memory area to a bounded buffer */
+static char*
+_bprint_hexdump( char* p, char* end, const uint8_t* data, int datalen )
+{
+ int lineSize = 16;
+
+ while (datalen > 0) {
+ int avail = datalen;
+ int nn;
+
+ if (avail > lineSize)
+ avail = lineSize;
+
+ for (nn = 0; nn < avail; nn++) {
+ if (nn > 0)
+ p = _bprint_c(p, end, ' ');
+ p = _bprint_hex(p, end, data[nn], 2);
+ }
+ for ( ; nn < lineSize; nn++ ) {
+ p = _bprint_s(p, end, " ");
+ }
+ p = _bprint_s(p, end, " ");
+
+ for (nn = 0; nn < avail; nn++) {
+ int c = data[nn];
+
+ if (c < 32 || c > 127)
+ c = '.';
+
+ p = _bprint_c(p, end, c);
+ }
+ p = _bprint_c(p, end, '\n');
+
+ data += avail;
+ datalen -= avail;
+ }
+ return p;
+}
+
+/* dump the content of a query of packet to the log */
+static void
+XLOG_BYTES( const void* base, int len )
+{
+ char buff[1024];
+ char* p = buff, *end = p + sizeof(buff);
+
+ p = _bprint_hexdump(p, end, base, len);
+ XLOG("%s",buff);
+}
+
+#else /* !DEBUG */
+# define XLOG(...) ((void)0)
+# define XLOG_BYTES(a,b) ((void)0)
+#endif
static time_t
_time_now( void )
@@ -107,96 +343,625 @@
return tv.tv_sec;
}
-/****************************************************************************/
-/****************************************************************************/
-/***** *****/
-/***** *****/
-/***** *****/
-/****************************************************************************/
-/****************************************************************************/
-
-/* used to define the content of _RESOLV_HOSTENT_NONE
+/* reminder: the general format of a DNS packet is the following:
+ *
+ * HEADER (12 bytes)
+ * QUESTION (variable)
+ * ANSWER (variable)
+ * AUTHORITY (variable)
+ * ADDITIONNAL (variable)
+ *
+ * the HEADER is made of:
+ *
+ * ID : 16 : 16-bit unique query identification field
+ *
+ * QR : 1 : set to 0 for queries, and 1 for responses
+ * Opcode : 4 : set to 0 for queries
+ * AA : 1 : set to 0 for queries
+ * TC : 1 : truncation flag, will be set to 0 in queries
+ * RD : 1 : recursion desired
+ *
+ * RA : 1 : recursion available (0 in queries)
+ * Z : 3 : three reserved zero bits
+ * RCODE : 4 : response code (always 0=NOERROR in queries)
+ *
+ * QDCount: 16 : question count
+ * ANCount: 16 : Answer count (0 in queries)
+ * NSCount: 16: Authority Record count (0 in queries)
+ * ARCount: 16: Additionnal Record count (0 in queries)
+ *
+ * the QUESTION is made of QDCount Question Record (QRs)
+ * the ANSWER is made of ANCount RRs
+ * the AUTHORITY is made of NSCount RRs
+ * the ADDITIONNAL is made of ARCount RRs
+ *
+ * Each Question Record (QR) is made of:
+ *
+ * QNAME : variable : Query DNS NAME
+ * TYPE : 16 : type of query (A=1, PTR=12, MX=15, AAAA=28, ALL=255)
+ * CLASS : 16 : class of query (IN=1)
+ *
+ * Each Resource Record (RR) is made of:
+ *
+ * NAME : variable : DNS NAME
+ * TYPE : 16 : type of query (A=1, PTR=12, MX=15, AAAA=28, ALL=255)
+ * CLASS : 16 : class of query (IN=1)
+ * TTL : 32 : seconds to cache this RR (0=none)
+ * RDLENGTH: 16 : size of RDDATA in bytes
+ * RDDATA : variable : RR data (depends on TYPE)
+ *
+ * Each QNAME contains a domain name encoded as a sequence of 'labels'
+ * terminated by a zero. Each label has the following format:
+ *
+ * LEN : 8 : lenght of label (MUST be < 64)
+ * NAME : 8*LEN : label length (must exclude dots)
+ *
+ * A value of 0 in the encoding is interpreted as the 'root' domain and
+ * terminates the encoding. So 'www.android.com' will be encoded as:
+ *
+ * <3>www<7>android<3>com<0>
+ *
+ * Where <n> represents the byte with value 'n'
+ *
+ * Each NAME reflects the QNAME of the question, but has a slightly more
+ * complex encoding in order to provide message compression. This is achieved
+ * by using a 2-byte pointer, with format:
+ *
+ * TYPE : 2 : 0b11 to indicate a pointer, 0b01 and 0b10 are reserved
+ * OFFSET : 14 : offset to another part of the DNS packet
+ *
+ * The offset is relative to the start of the DNS packet and must point
+ * A pointer terminates the encoding.
+ *
+ * The NAME can be encoded in one of the following formats:
+ *
+ * - a sequence of simple labels terminated by 0 (like QNAMEs)
+ * - a single pointer
+ * - a sequence of simple labels terminated by a pointer
+ *
+ * A pointer shall always point to either a pointer of a sequence of
+ * labels (which can themselves be terminated by either a 0 or a pointer)
+ *
+ * The expanded length of a given domain name should not exceed 255 bytes.
+ *
+ * NOTE: we don't parse the answer packets, so don't need to deal with NAME
+ * records, only QNAMEs.
*/
-const struct hostent _resolv_hostent_none_cst = {
- NULL,
- NULL,
- AF_INET,
- 4,
- NULL
-};
-struct hostent*
-_resolv_hostent_copy( struct hostent* hp )
+#define DNS_HEADER_SIZE 12
+
+#define DNS_TYPE_A "\00\01" /* big-endian decimal 1 */
+#define DNS_TYPE_PTR "\00\014" /* big-endian decimal 12 */
+#define DNS_TYPE_MX "\00\017" /* big-endian decimal 15 */
+#define DNS_TYPE_AAAA "\00\034" /* big-endian decimal 28 */
+#define DNS_TYPE_ALL "\00\0377" /* big-endian decimal 255 */
+
+#define DNS_CLASS_IN "\00\01" /* big-endian decimal 1 */
+
+typedef struct {
+ const uint8_t* base;
+ const uint8_t* end;
+ const uint8_t* cursor;
+} DnsPacket;
+
+static void
+_dnsPacket_init( DnsPacket* packet, const uint8_t* buff, int bufflen )
{
- struct hostent* dst;
- int nn, len;
- char* p;
- int num_aliases = 0, num_addresses = 0;
+ packet->base = buff;
+ packet->end = buff + bufflen;
+ packet->cursor = buff;
+}
- if (hp == NULL)
- return NULL;
+static void
+_dnsPacket_rewind( DnsPacket* packet )
+{
+ packet->cursor = packet->base;
+}
- if (hp == _RESOLV_HOSTENT_NONE)
- return hp;
+static void
+_dnsPacket_skip( DnsPacket* packet, int count )
+{
+ const uint8_t* p = packet->cursor + count;
- len = sizeof(*hp);
- len += strlen(hp->h_name) + 1;
+ if (p > packet->end)
+ p = packet->end;
- if (hp->h_aliases != NULL) {
- for (nn = 0; hp->h_aliases[nn] != NULL; nn++)
- len += sizeof(char*) + strlen(hp->h_aliases[nn]) + 1;
- num_aliases = nn;
+ packet->cursor = p;
+}
+
+static int
+_dnsPacket_readInt16( DnsPacket* packet )
+{
+ const uint8_t* p = packet->cursor;
+
+ if (p+2 > packet->end)
+ return -1;
+
+ packet->cursor = p+2;
+ return (p[0]<< 8) | p[1];
+}
+
+/** QUERY CHECKING
+ **/
+
+/* check bytes in a dns packet. returns 1 on success, 0 on failure.
+ * the cursor is only advanced in the case of success
+ */
+static int
+_dnsPacket_checkBytes( DnsPacket* packet, int numBytes, const void* bytes )
+{
+ const uint8_t* p = packet->cursor;
+
+ if (p + numBytes > packet->end)
+ return 0;
+
+ if (memcmp(p, bytes, numBytes) != 0)
+ return 0;
+
+ packet->cursor = p + numBytes;
+ return 1;
+}
+
+/* parse and skip a given QNAME stored in a query packet,
+ * from the current cursor position. returns 1 on success,
+ * or 0 for malformed data.
+ */
+static int
+_dnsPacket_checkQName( DnsPacket* packet )
+{
+ const uint8_t* p = packet->cursor;
+ const uint8_t* end = packet->end;
+
+ for (;;) {
+ int c;
+
+ if (p >= end)
+ break;
+
+ c = *p++;
+
+ if (c == 0) {
+ packet->cursor = p;
+ return 1;
+ }
+
+ /* we don't expect label compression in QNAMEs */
+ if (c >= 64)
+ break;
+
+ p += c;
+ /* we rely on the bound check at the start
+ * of the loop here */
}
- len += sizeof(char*);
+ /* malformed data */
+ XLOG("malformed QNAME");
+ return 0;
+}
- for (nn = 0; hp->h_addr_list[nn] != NULL; nn++) {
- len += sizeof(char*) + hp->h_length;
+/* parse and skip a given QR stored in a packet.
+ * returns 1 on success, and 0 on failure
+ */
+static int
+_dnsPacket_checkQR( DnsPacket* packet )
+{
+ int len;
+
+ if (!_dnsPacket_checkQName(packet))
+ return 0;
+
+ /* TYPE must be one of the things we support */
+ if (!_dnsPacket_checkBytes(packet, 2, DNS_TYPE_A) &&
+ !_dnsPacket_checkBytes(packet, 2, DNS_TYPE_PTR) &&
+ !_dnsPacket_checkBytes(packet, 2, DNS_TYPE_MX) &&
+ !_dnsPacket_checkBytes(packet, 2, DNS_TYPE_AAAA) &&
+ !_dnsPacket_checkBytes(packet, 2, DNS_TYPE_ALL))
+ {
+ XLOG("unsupported TYPE");
+ return 0;
}
- num_addresses = nn;
- len += sizeof(char*);
-
- dst = malloc( len );
- if (dst == NULL)
- return NULL;
-
- dst->h_aliases = (char**)(dst + 1);
- dst->h_addr_list = dst->h_aliases + num_aliases + 1;
- dst->h_length = hp->h_length;
- dst->h_addrtype = hp->h_addrtype;
-
- p = (char*)(dst->h_addr_list + num_addresses + 1);
-
- /* write the addresses first, to help with alignment issues */
- for (nn = 0; nn < num_addresses; nn++) {
- dst->h_addr_list[nn] = p;
- len = hp->h_length;
- memcpy( p, hp->h_addr_list[nn], len );
- p += len;
+ /* CLASS must be IN */
+ if (!_dnsPacket_checkBytes(packet, 2, DNS_CLASS_IN)) {
+ XLOG("unsupported CLASS");
+ return 0;
}
- dst->h_addr_list[nn] = NULL;
- for (nn = 0; nn < num_aliases; nn++) {
- dst->h_aliases[nn] = p;
- len = strlen(hp->h_aliases[nn]) + 1;
- memcpy( p, hp->h_aliases[nn], len );
- p += len;
+ return 1;
+}
+
+/* check the header of a DNS Query packet, return 1 if it is one
+ * type of query we can cache, or 0 otherwise
+ */
+static int
+_dnsPacket_checkQuery( DnsPacket* packet )
+{
+ const uint8_t* p = packet->base;
+ int qdCount, anCount, dnCount, arCount;
+
+ if (p + DNS_HEADER_SIZE > packet->end) {
+ XLOG("query packet too small");
+ return 0;
}
- dst->h_aliases[nn] = NULL;
- dst->h_name = p;
- len = strlen(hp->h_name) + 1;
- memcpy(p, hp->h_name, len);
- p += len;
+ /* QR must be set to 0, opcode must be 0 and AA must be 0 */
+ /* RA, Z, and RCODE must be 0 */
+ if ((p[2] & 0xFC) != 0 || p[3] != 0) {
+ XLOG("query packet flags unsupported");
+ return 0;
+ }
- return dst;
+ /* Note that we ignore the TC and RD bits here for the
+ * following reasons:
+ *
+ * - there is no point for a query packet sent to a server
+ * to have the TC bit set, but the implementation might
+ * set the bit in the query buffer for its own needs
+ * between a _resolv_cache_lookup and a
+ * _resolv_cache_add. We should not freak out if this
+ * is the case.
+ *
+ * - we consider that the result from a RD=0 or a RD=1
+ * query might be different, hence that the RD bit
+ * should be used to differentiate cached result.
+ *
+ * this implies that RD is checked when hashing or
+ * comparing query packets, but not TC
+ */
+
+ /* ANCOUNT, DNCOUNT and ARCOUNT must be 0 */
+ qdCount = (p[4] << 8) | p[5];
+ anCount = (p[6] << 8) | p[7];
+ dnCount = (p[8] << 8) | p[9];
+ arCount = (p[10]<< 8) | p[11];
+
+ if (anCount != 0 || dnCount != 0 || arCount != 0) {
+ XLOG("query packet contains non-query records");
+ return 0;
+ }
+
+ if (qdCount == 0) {
+ XLOG("query packet doesn't contain query record");
+ return 0;
+ }
+
+ /* Check QDCOUNT QRs */
+ packet->cursor = p + DNS_HEADER_SIZE;
+
+ for (;qdCount > 0; qdCount--)
+ if (!_dnsPacket_checkQR(packet))
+ return 0;
+
+ return 1;
+}
+
+/** QUERY DEBUGGING
+ **/
+#if DEBUG
+static char*
+_dnsPacket_bprintQName(DnsPacket* packet, char* bp, char* bend)
+{
+ const uint8_t* p = packet->cursor;
+ const uint8_t* end = packet->end;
+ int first = 1;
+
+ for (;;) {
+ int c;
+
+ if (p >= end)
+ break;
+
+ c = *p++;
+
+ if (c == 0) {
+ packet->cursor = p;
+ return bp;
+ }
+
+ /* we don't expect label compression in QNAMEs */
+ if (c >= 64)
+ break;
+
+ if (first)
+ first = 0;
+ else
+ bp = _bprint_c(bp, bend, '.');
+
+ bp = _bprint_b(bp, bend, (const char*)p, c);
+
+ p += c;
+ /* we rely on the bound check at the start
+ * of the loop here */
+ }
+ /* malformed data */
+ bp = _bprint_s(bp, bend, "<MALFORMED>");
+ return bp;
+}
+
+static char*
+_dnsPacket_bprintQR(DnsPacket* packet, char* p, char* end)
+{
+#define QQ(x) { DNS_TYPE_##x, #x }
+ static const struct {
+ const char* typeBytes;
+ const char* typeString;
+ } qTypes[] =
+ {
+ QQ(A), QQ(PTR), QQ(MX), QQ(AAAA), QQ(ALL),
+ { NULL, NULL }
+ };
+ int nn;
+ const char* typeString = NULL;
+
+ /* dump QNAME */
+ p = _dnsPacket_bprintQName(packet, p, end);
+
+ /* dump TYPE */
+ p = _bprint_s(p, end, " (");
+
+ for (nn = 0; qTypes[nn].typeBytes != NULL; nn++) {
+ if (_dnsPacket_checkBytes(packet, 2, qTypes[nn].typeBytes)) {
+ typeString = qTypes[nn].typeString;
+ break;
+ }
+ }
+
+ if (typeString != NULL)
+ p = _bprint_s(p, end, typeString);
+ else {
+ int typeCode = _dnsPacket_readInt16(packet);
+ p = _bprint(p, end, "UNKNOWN-%d", typeCode);
+ }
+
+ p = _bprint_c(p, end, ')');
+
+ /* skip CLASS */
+ _dnsPacket_skip(packet, 2);
+ return p;
+}
+
+/* this function assumes the packet has already been checked */
+static char*
+_dnsPacket_bprintQuery( DnsPacket* packet, char* p, char* end )
+{
+ int qdCount;
+
+ if (packet->base[2] & 0x1) {
+ p = _bprint_s(p, end, "RECURSIVE ");
+ }
+
+ _dnsPacket_skip(packet, 4);
+ qdCount = _dnsPacket_readInt16(packet);
+ _dnsPacket_skip(packet, 6);
+
+ for ( ; qdCount > 0; qdCount-- ) {
+ p = _dnsPacket_bprintQR(packet, p, end);
+ }
+ return p;
+}
+#endif
+
+
+/** QUERY HASHING SUPPORT
+ **
+ ** THE FOLLOWING CODE ASSUMES THAT THE INPUT PACKET HAS ALREADY
+ ** BEEN SUCCESFULLY CHECKED.
+ **/
+
+/* use 32-bit FNV hash function */
+#define FNV_MULT 16777619U
+#define FNV_BASIS 2166136261U
+
+static unsigned
+_dnsPacket_hashBytes( DnsPacket* packet, int numBytes, unsigned hash )
+{
+ const uint8_t* p = packet->cursor;
+ const uint8_t* end = packet->end;
+
+ while (numBytes > 0 && p < end) {
+ hash = hash*FNV_MULT ^ *p++;
+ }
+ packet->cursor = p;
+ return hash;
}
-void
-_resolv_hostent_free( struct hostent* hp )
+static unsigned
+_dnsPacket_hashQName( DnsPacket* packet, unsigned hash )
{
- if (hp && hp != _RESOLV_HOSTENT_NONE)
- free(hp);
+ const uint8_t* p = packet->cursor;
+ const uint8_t* end = packet->end;
+
+ for (;;) {
+ int c;
+
+ if (p >= end) { /* should not happen */
+ XLOG("%s: INTERNAL_ERROR: read-overflow !!\n", __FUNCTION__);
+ break;
+ }
+
+ c = *p++;
+
+ if (c == 0)
+ break;
+
+ if (c >= 64) {
+ XLOG("%s: INTERNAL_ERROR: malformed domain !!\n", __FUNCTION__);
+ break;
+ }
+ if (p + c >= end) {
+ XLOG("%s: INTERNAL_ERROR: simple label read-overflow !!\n",
+ __FUNCTION__);
+ break;
+ }
+ while (c > 0) {
+ hash = hash*FNV_MULT ^ *p++;
+ c -= 1;
+ }
+ }
+ packet->cursor = p;
+ return hash;
+}
+
+static unsigned
+_dnsPacket_hashQR( DnsPacket* packet, unsigned hash )
+{
+ int len;
+
+ hash = _dnsPacket_hashQName(packet, hash);
+ hash = _dnsPacket_hashBytes(packet, 4, hash); /* TYPE and CLASS */
+ return hash;
+}
+
+static unsigned
+_dnsPacket_hashQuery( DnsPacket* packet )
+{
+ unsigned hash = FNV_BASIS;
+ int count;
+ _dnsPacket_rewind(packet);
+
+ /* we ignore the TC bit for reasons explained in
+ * _dnsPacket_checkQuery().
+ *
+ * however we hash the RD bit to differentiate
+ * between answers for recursive and non-recursive
+ * queries.
+ */
+ hash = hash*FNV_MULT ^ (packet->base[2] & 1);
+
+ /* assume: other flags are 0 */
+ _dnsPacket_skip(packet, 4);
+
+ /* read QDCOUNT */
+ count = _dnsPacket_readInt16(packet);
+
+ /* assume: ANcount, NScount, ARcount are 0 */
+ _dnsPacket_skip(packet, 6);
+
+ /* hash QDCOUNT QRs */
+ for ( ; count > 0; count-- )
+ hash = _dnsPacket_hashQR(packet, hash);
+
+ return hash;
+}
+
+
+/** QUERY COMPARISON
+ **
+ ** THE FOLLOWING CODE ASSUMES THAT THE INPUT PACKETS HAVE ALREADY
+ ** BEEN SUCCESFULLY CHECKED.
+ **/
+
+static int
+_dnsPacket_isEqualDomainName( DnsPacket* pack1, DnsPacket* pack2 )
+{
+ const uint8_t* p1 = pack1->cursor;
+ const uint8_t* end1 = pack1->end;
+ const uint8_t* p2 = pack2->cursor;
+ const uint8_t* end2 = pack2->end;
+
+ for (;;) {
+ int c1, c2;
+
+ if (p1 >= end1 || p2 >= end2) {
+ XLOG("%s: INTERNAL_ERROR: read-overflow !!\n", __FUNCTION__);
+ break;
+ }
+ c1 = *p1++;
+ c2 = *p2++;
+ if (c1 != c2)
+ break;
+
+ if (c1 == 0) {
+ pack1->cursor = p1;
+ pack2->cursor = p2;
+ return 1;
+ }
+ if (c1 >= 64) {
+ XLOG("%s: INTERNAL_ERROR: malformed domain !!\n", __FUNCTION__);
+ break;
+ }
+ if ((p1+c1 > end1) || (p2+c1 > end2)) {
+ XLOG("%s: INTERNAL_ERROR: simple label read-overflow !!\n",
+ __FUNCTION__);
+ break;
+ }
+ if (memcmp(p1, p2, c1) != 0)
+ break;
+ p1 += c1;
+ p2 += c1;
+ /* we rely on the bound checks at the start of the loop */
+ }
+ /* not the same, or one is malformed */
+ XLOG("different DN");
+ return 0;
+}
+
+static int
+_dnsPacket_isEqualBytes( DnsPacket* pack1, DnsPacket* pack2, int numBytes )
+{
+ const uint8_t* p1 = pack1->cursor;
+ const uint8_t* p2 = pack2->cursor;
+
+ if ( p1 + numBytes > pack1->end || p2 + numBytes > pack2->end )
+ return 0;
+
+ if ( memcmp(p1, p2, numBytes) != 0 )
+ return 0;
+
+ pack1->cursor += numBytes;
+ pack2->cursor += numBytes;
+ return 1;
+}
+
+static int
+_dnsPacket_isEqualQR( DnsPacket* pack1, DnsPacket* pack2 )
+{
+ /* compare domain name encoding + TYPE + CLASS */
+ if ( !_dnsPacket_isEqualDomainName(pack1, pack2) ||
+ !_dnsPacket_isEqualBytes(pack1, pack2, 2+2) )
+ return 0;
+
+ return 1;
+}
+
+static int
+_dnsPacket_isEqualQuery( DnsPacket* pack1, DnsPacket* pack2 )
+{
+ int count1, count2;
+
+ /* compare the headers, ignore most fields */
+ _dnsPacket_rewind(pack1);
+ _dnsPacket_rewind(pack2);
+
+ /* compare RD, ignore TC, see comment in _dnsPacket_checkQuery */
+ if ((pack1->base[2] & 1) != (pack2->base[2] & 1)) {
+ XLOG("different RD");
+ return 0;
+ }
+
+ /* assume: other flags are all 0 */
+ _dnsPacket_skip(pack1, 4);
+ _dnsPacket_skip(pack2, 4);
+
+ /* compare QDCOUNT */
+ count1 = _dnsPacket_readInt16(pack1);
+ count2 = _dnsPacket_readInt16(pack2);
+ if (count1 != count2 || count1 < 0) {
+ XLOG("different QDCOUNT");
+ return 0;
+ }
+
+ /* assume: ANcount, NScount and ARcount are all 0 */
+ _dnsPacket_skip(pack1, 6);
+ _dnsPacket_skip(pack2, 6);
+
+ /* compare the QDCOUNT QRs */
+ for ( ; count1 > 0; count1-- ) {
+ if (!_dnsPacket_isEqualQR(pack1, pack2)) {
+ XLOG("different QR");
+ return 0;
+ }
+ }
+ return 1;
}
/****************************************************************************/
@@ -207,15 +972,23 @@
/****************************************************************************/
/****************************************************************************/
+/* cache entry. for simplicity, 'hash' and 'hlink' are inlined in this
+ * structure though they are conceptually part of the hash table.
+ *
+ * similarly, mru_next and mru_prev are part of the global MRU list
+ */
typedef struct Entry {
- unsigned int hash;
- const char* name;
- short af;
- short index;
+ unsigned int hash; /* hash value */
+ struct Entry* hlink; /* next in collision chain */
struct Entry* mru_prev;
struct Entry* mru_next;
- time_t when;
- struct hostent* hp;
+
+ const uint8_t* query;
+ int querylen;
+ const uint8_t* answer;
+ int answerlen;
+ time_t when; /* time_t when entry was added to table */
+ int id; /* for debugging purpose */
} Entry;
@@ -224,30 +997,10 @@
{
/* everything is allocated in a single memory block */
if (e) {
- _resolv_hostent_free(e->hp);
free(e);
}
}
-static void
-entry_init_key( Entry* e, const char* name, int af )
-{
- unsigned h = 0;
- const char* p = name;
-
- /* compute hash */
- p = name;
- while (*p) {
- h = h*33 + *p++;
- }
- h += af*17;
-
- e->hash = h;
- e->name = name;
- e->af = (short) af;
-}
-
-
static __inline__ void
entry_mru_remove( Entry* e )
{
@@ -267,46 +1020,75 @@
first->mru_prev = e;
}
+/* compute the hash of a given entry, this is a hash of most
+ * data in the query (key) */
+static unsigned
+entry_hash( const Entry* e )
+{
+ DnsPacket pack[1];
+ _dnsPacket_init(pack, e->query, e->querylen);
+ return _dnsPacket_hashQuery(pack);
+}
+
+/* initialize an Entry as a search key, this also checks the input query packet
+ * returns 1 on success, or 0 in case of unsupported/malformed data */
+static int
+entry_init_key( Entry* e, const void* query, int querylen )
+{
+ DnsPacket pack[1];
+
+ memset(e, 0, sizeof(*e));
+
+ e->query = query;
+ e->querylen = querylen;
+ e->hash = entry_hash(e);
+
+ _dnsPacket_init(pack, query, querylen);
+
+ return _dnsPacket_checkQuery(pack);
+}
+
+/* allocate a new entry as a cache node */
static Entry*
-entry_alloc( const char* name, int af, int index, struct hostent* hp )
+entry_alloc( const Entry* init, const void* answer, int answerlen )
{
Entry* e;
- int num_aliases = 0;
- int num_addresses = 0;
- char** aliases;
- char** addresses;
+ int size;
- /* compute the length of the memory block that will contain everything */
- int len = sizeof(*e) + strlen(name)+1;
-
- e = malloc(len);
+ size = sizeof(*e) + init->querylen + answerlen;
+ e = calloc(size, 1);
if (e == NULL)
return e;
- entry_init_key(e, name, af);
+ e->hash = init->hash;
+ e->query = (const uint8_t*)(e+1);
+ e->querylen = init->querylen;
- e->mru_next = e->mru_prev = e;
- e->index = (short) index;
- e->when = _time_now();
- e->hp = _resolv_hostent_copy(hp);
+ memcpy( (char*)e->query, init->query, e->querylen );
- if (e->hp == NULL) {
- free(e);
- return NULL;
- }
+ e->answer = e->query + e->querylen;
+ e->answerlen = answerlen;
- e->name = (char*)(e+1);
- len = strlen(name)+1;
- memcpy( (char*)e->name, name, len );
+ memcpy( (char*)e->answer, answer, e->answerlen );
+
+ e->when = _time_now();
+
return e;
}
-
-static __inline__ int
+static int
entry_equals( const Entry* e1, const Entry* e2 )
{
- return e1->hash == e2->hash && e1->af == e2->af && !strcmp( e1->name, e2->name );
+ DnsPacket pack1[1], pack2[1];
+
+ if (e1->querylen != e2->querylen) {
+ return 0;
+ }
+ _dnsPacket_init(pack1, e1->query, e1->querylen);
+ _dnsPacket_init(pack2, e2->query, e2->querylen);
+
+ return _dnsPacket_isEqualQuery(pack1, pack2);
}
/****************************************************************************/
@@ -317,30 +1099,49 @@
/****************************************************************************/
/****************************************************************************/
+/* We use a simple hash table with external collision lists
+ * for simplicity, the hash-table fields 'hash' and 'hlink' are
+ * inlined in the Entry structure.
+ */
#define MAX_HASH_ENTRIES (2*CONFIG_MAX_ENTRIES)
typedef struct resolv_cache {
- int num_entries;
- Entry mru_list;
- pthread_mutex_t lock;
- int disabled;
- Entry* entries[ MAX_HASH_ENTRIES ]; /* hash-table of pointers to entries */
+ int num_entries;
+ Entry mru_list;
+ pthread_mutex_t lock;
+ unsigned generation;
+ int last_id;
+ Entry* entries[ MAX_HASH_ENTRIES ];
} Cache;
-void
-_resolv_cache_destroy( struct resolv_cache* cache )
-{
- if (cache != NULL) {
- int nn;
- for (nn = 0; nn < MAX_HASH_ENTRIES; nn++) {
- entry_free(cache->entries[nn]);
- }
- pthread_mutex_destroy(&cache->lock);
- free(cache);
- }
-}
+#define HTABLE_VALID(x) ((x) != NULL && (x) != HTABLE_DELETED)
+static void
+_cache_flush_locked( Cache* cache )
+{
+ int nn;
+ time_t now = _time_now();
+
+ for (nn = 0; nn < MAX_HASH_ENTRIES; nn++)
+ {
+ Entry** pnode = &cache->entries[nn];
+
+ while (*pnode != NULL) {
+ Entry* node = *pnode;
+ *pnode = node->hlink;
+ entry_free(node);
+ }
+ }
+
+ cache->mru_list.mru_next = cache->mru_list.mru_prev = &cache->mru_list;
+ cache->num_entries = 0;
+ cache->last_id = 0;
+
+ XLOG("*************************\n"
+ "*** DNS CACHE FLUSHED ***\n"
+ "*************************");
+}
struct resolv_cache*
_resolv_cache_create( void )
@@ -349,116 +1150,203 @@
cache = calloc(sizeof(*cache), 1);
if (cache) {
- const char* env = getenv(CONFIG_ENV);
-
- if (env && atoi(env) == 0)
- cache->disabled = 1;
-
+ cache->generation = ~0U;
pthread_mutex_init( &cache->lock, NULL );
cache->mru_list.mru_prev = cache->mru_list.mru_next = &cache->mru_list;
- XLOG("%s: cache=%p %s\n", __FUNCTION__, cache, cache->disabled ? "disabled" : "enabled" );
+ XLOG("%s: cache created\n", __FUNCTION__);
}
return cache;
}
-static int
-_resolv_cache_find_index( Cache* cache,
- const char* name,
- int af )
+#if DEBUG
+static void
+_dump_query( const uint8_t* query, int querylen )
{
- Entry key;
- int nn, step, tries;
+ char temp[256], *p=temp, *end=p+sizeof(temp);
+ DnsPacket pack[1];
- entry_init_key( &key, name, af );
-
- tries = MAX_HASH_ENTRIES;
- nn = key.hash % MAX_HASH_ENTRIES;
- step = 5;
-
- while (tries > 0) {
- Entry* key2 = cache->entries[nn];
-
- if (key2 == NULL) {
- return -(nn + 1);
- }
-
- if (entry_equals( &key, key2 ) ) {
- return nn;
- }
-
- nn = (nn + step) % MAX_HASH_ENTRIES;
- tries -= 1;
- }
- return -(MAX_HASH_ENTRIES+1);
+ _dnsPacket_init(pack, query, querylen);
+ p = _dnsPacket_bprintQuery(pack, p, end);
+ XLOG("QUERY: %s", temp);
}
-
static void
-_resolv_cache_remove( struct resolv_cache* cache,
- Entry* e )
+_cache_dump_mru( Cache* cache )
{
- XLOG("%s: name='%s' af=%d\n", __FUNCTION__, e->name, e->af);
- cache->entries[ e->index ] = NULL; /* remove from hash table */
- entry_mru_remove( e );
- entry_free( e );
+ char temp[512], *p=temp, *end=p+sizeof(temp);
+ Entry* e;
+
+ p = _bprint(temp, end, "MRU LIST (%2d): ", cache->num_entries);
+ for (e = cache->mru_list.mru_next; e != &cache->mru_list; e = e->mru_next)
+ p = _bprint(p, end, " %d", e->id);
+
+ XLOG("%s", temp);
+}
+#endif
+
+#if DEBUG
+# define XLOG_QUERY(q,len) _dump_query((q), (len))
+#else
+# define XLOG_QUERY(q,len) ((void)0)
+#endif
+
+/* This function tries to find a key within the hash table
+ * In case of success, it will return a *pointer* to the hashed key.
+ * In case of failure, it will return a *pointer* to NULL
+ *
+ * So, the caller must check '*result' to check for success/failure.
+ *
+ * The main idea is that the result can later be used directly in
+ * calls to _resolv_cache_add or _resolv_cache_remove as the 'lookup'
+ * parameter. This makes the code simpler and avoids re-searching
+ * for the key position in the htable.
+ *
+ * The result of a lookup_p is only valid until you alter the hash
+ * table.
+ */
+static Entry**
+_cache_lookup_p( Cache* cache,
+ Entry* key )
+{
+ int index = key->hash % MAX_HASH_ENTRIES;
+ Entry** pnode = &cache->entries[ key->hash % MAX_HASH_ENTRIES ];
+
+ while (*pnode != NULL) {
+ Entry* node = *pnode;
+
+ if (node == NULL)
+ break;
+
+ if (node->hash == key->hash && entry_equals(node, key))
+ break;
+
+ pnode = &node->hlink;
+ }
+ return pnode;
+}
+
+/* Add a new entry to the hash table. 'lookup' must be the
+ * result of an immediate previous failed _lookup_p() call
+ * (i.e. with *lookup == NULL), and 'e' is the pointer to the
+ * newly created entry
+ */
+static void
+_cache_add_p( Cache* cache,
+ Entry** lookup,
+ Entry* e )
+{
+ *lookup = e;
+ e->id = ++cache->last_id;
+ entry_mru_add(e, &cache->mru_list);
+ cache->num_entries += 1;
+
+ XLOG("%s: entry %d added (count=%d)", __FUNCTION__,
+ e->id, cache->num_entries);
+}
+
+/* Remove an existing entry from the hash table,
+ * 'lookup' must be the result of an immediate previous
+ * and succesful _lookup_p() call.
+ */
+static void
+_cache_remove_p( Cache* cache,
+ Entry** lookup )
+{
+ Entry* e = *lookup;
+
+ XLOG("%s: entry %d removed (count=%d)", __FUNCTION__,
+ e->id, cache->num_entries-1);
+
+ entry_mru_remove(e);
+ *lookup = e->hlink;
+ entry_free(e);
cache->num_entries -= 1;
}
-
-struct hostent*
-_resolv_cache_lookup( struct resolv_cache* cache,
- const char* name,
- int af )
+/* Remove the oldest entry from the hash table.
+ */
+static void
+_cache_remove_oldest( Cache* cache )
{
- int index;
- struct hostent* result = NULL;
+ Entry* oldest = cache->mru_list.mru_prev;
+ Entry** lookup = _cache_lookup_p(cache, oldest);
- if (cache->disabled)
- return NULL;
+ if (*lookup == NULL) { /* should not happen */
+ XLOG("%s: OLDEST NOT IN HTABLE ?", __FUNCTION__);
+ return;
+ }
+ _cache_remove_p(cache, lookup);
+}
+
+ResolvCacheStatus
+_resolv_cache_lookup( struct resolv_cache* cache,
+ const void* query,
+ int querylen,
+ void* answer,
+ int answersize,
+ int *answerlen )
+{
+ DnsPacket pack[1];
+ Entry key[1];
+ int index;
+ Entry** lookup;
+ Entry* e;
+ time_t now;
+
+ ResolvCacheStatus result = RESOLV_CACHE_NOTFOUND;
+
+ XLOG("%s: lookup", __FUNCTION__);
+ XLOG_QUERY(query, querylen);
+
+ /* we don't cache malformed queries */
+ if (!entry_init_key(key, query, querylen)) {
+ XLOG("%s: unsupported query", __FUNCTION__);
+ return RESOLV_CACHE_UNSUPPORTED;
+ }
+ /* lookup cache */
pthread_mutex_lock( &cache->lock );
- XLOG( "%s: cache=%p name='%s' af=%d ", __FUNCTION__, cache, name, af );
- index = _resolv_cache_find_index( cache, name, af );
- if (index >= 0) {
- Entry* e = cache->entries[index];
- time_t now = _time_now();
- struct hostent** pht;
+ /* see the description of _lookup_p to understand this.
+ * the function always return a non-NULL pointer.
+ */
+ lookup = _cache_lookup_p(cache, key);
+ e = *lookup;
- /* ignore stale entries, they will be discarded in _resolv_cache_add */
- if ( (unsigned)(now - e->when) >= CONFIG_SECONDS ) {
- XLOG( " OLD\n" );
- goto Exit;
- }
-
- /* bump up this entry to the top of the MRU list */
- if (e != cache->mru_list.mru_next) {
- entry_mru_remove( e );
- entry_mru_add( e, &cache->mru_list );
- }
-
- /* now copy the result into a thread-local variable */
- pht = __get_res_cache_hostent_p();
- if (pht == NULL) {
- XLOG( " NOTLS\n" ); /* shouldn't happen */
- goto Exit;
- }
-
- if (pht[0]) {
- _resolv_hostent_free( pht[0] ); /* clear previous value */
- pht[0] = NULL;
- }
- result = _resolv_hostent_copy( e->hp );
- if (result == NULL) {
- XLOG( " NOMEM\n" ); /* bummer */
- goto Exit;
- }
- XLOG( " OK\n" );
- pht[0] = result;
+ if (e == NULL) {
+ XLOG( "NOT IN CACHE");
goto Exit;
}
- XLOG( " KO\n" );
+
+ now = _time_now();
+
+ /* remove stale entries here */
+ if ( (unsigned)(now - e->when) >= CONFIG_SECONDS ) {
+ XLOG( " NOT IN CACHE (STALE ENTRY %p DISCARDED)", *lookup );
+ _cache_remove_p(cache, lookup);
+ goto Exit;
+ }
+
+ *answerlen = e->answerlen;
+ if (e->answerlen > answersize) {
+ /* NOTE: we return UNSUPPORTED if the answer buffer is too short */
+ result = RESOLV_CACHE_UNSUPPORTED;
+ XLOG(" ANSWER TOO LONG");
+ goto Exit;
+ }
+
+ memcpy( answer, e->answer, e->answerlen );
+
+ /* bump up this entry to the top of the MRU list */
+ if (e != cache->mru_list.mru_next) {
+ entry_mru_remove( e );
+ entry_mru_add( e, &cache->mru_list );
+ }
+
+ XLOG( "FOUND IN CACHE entry=%p", e );
+ result = RESOLV_CACHE_FOUND;
+
Exit:
pthread_mutex_unlock( &cache->lock );
return result;
@@ -467,42 +1355,59 @@
void
_resolv_cache_add( struct resolv_cache* cache,
- const char* name,
- int af,
- struct hostent* hp )
+ const void* query,
+ int querylen,
+ const void* answer,
+ int answerlen )
{
+ Entry key[1];
Entry* e;
- int index;
+ Entry** lookup;
- if (cache->disabled)
+ /* don't assume that the query has already been cached
+ */
+ if (!entry_init_key( key, query, querylen )) {
+ XLOG( "%s: passed invalid query ?", __FUNCTION__);
return;
+ }
pthread_mutex_lock( &cache->lock );
- XLOG( "%s: cache=%p name='%s' af=%d\n", __FUNCTION__, cache, name, af);
+ XLOG( "%s: query:", __FUNCTION__ );
+ XLOG_QUERY(query,querylen);
+#if DEBUG_DATA
+ XLOG( "answer:");
+ XLOG_BYTES(answer,answerlen);
+#endif
- /* get rid of the oldest entry if needed */
+ lookup = _cache_lookup_p(cache, key);
+ e = *lookup;
+
+ if (e != NULL) { /* should not happen */
+ XLOG("%s: ALREADY IN CACHE (%p) ? IGNORING ADD",
+ __FUNCTION__, e);
+ goto Exit;
+ }
+
if (cache->num_entries >= CONFIG_MAX_ENTRIES) {
- Entry* oldest = cache->mru_list.mru_prev;
- _resolv_cache_remove( cache, oldest );
+ _cache_remove_oldest(cache);
+ /* need to lookup again */
+ lookup = _cache_lookup_p(cache, key);
+ e = *lookup;
+ if (e != NULL) {
+ XLOG("%s: ALREADY IN CACHE (%p) ? IGNORING ADD",
+ __FUNCTION__, e);
+ goto Exit;
+ }
}
- index = _resolv_cache_find_index( cache, name, af );
- if (index >= 0) {
- /* discard stale entry */
- _resolv_cache_remove( cache, cache->entries[index] );
- } else {
- index = -(index+1);
- if (index >= MAX_HASH_ENTRIES)
- goto Exit; /* should not happen */
- }
-
- e = entry_alloc( name, af, index, hp );
+ e = entry_alloc( key, answer, answerlen );
if (e != NULL) {
- entry_mru_add( e, &cache->mru_list );
- cache->entries[index] = e;
- cache->num_entries += 1;
+ _cache_add_p(cache, lookup, e);
}
+#if DEBUG
+ _cache_dump_mru(cache);
+#endif
Exit:
pthread_mutex_unlock( &cache->lock );
}
@@ -521,6 +1426,13 @@
static void
_res_cache_init( void )
{
+ const char* env = getenv(CONFIG_ENV);
+
+ if (env && atoi(env) == 0) {
+ /* the cache is disabled */
+ return;
+ }
+
_res_cache = _resolv_cache_create();
}
@@ -531,3 +1443,19 @@
pthread_once( &_res_cache_once, _res_cache_init );
return _res_cache;
}
+
+void
+_resolv_cache_reset( unsigned generation )
+{
+ XLOG("%s: generation=%d", __FUNCTION__, generation);
+
+ if (_res_cache == NULL)
+ return;
+
+ pthread_mutex_lock( &_res_cache->lock );
+ if (_res_cache->generation != generation) {
+ _cache_flush_locked(_res_cache);
+ _res_cache->generation = generation;
+ }
+ pthread_mutex_unlock( &_res_cache->lock );
+}
diff --git a/libc/netbsd/resolv/res_send.c b/libc/netbsd/resolv/res_send.c
index 24b740a..3aca760 100644
--- a/libc/netbsd/resolv/res_send.c
+++ b/libc/netbsd/resolv/res_send.c
@@ -81,6 +81,9 @@
#endif
#endif /* LIBC_SCCS and not lint */
+/* set to 1 to use our small/simple/limited DNS cache */
+#define USE_RESOLV_CACHE 1
+
/*
* Send query to name server and wait for reply.
*/
@@ -111,6 +114,10 @@
#include <isc/eventlib.h>
+#if USE_RESOLV_CACHE
+# include <resolv_cache.h>
+#endif
+
#ifndef DE_CONST
#define DE_CONST(c,v) v = ((c) ? \
strchr((const void *)(c), *(const char *)(const void *)(c)) : NULL)
@@ -344,12 +351,17 @@
return (1);
}
+
int
res_nsend(res_state statp,
const u_char *buf, int buflen, u_char *ans, int anssiz)
{
int gotsomewhere, terrno, try, v_circuit, resplen, ns, n;
char abuf[NI_MAXHOST];
+#if USE_RESOLV_CACHE
+ struct resolv_cache* cache;
+ ResolvCacheStatus cache_status = RESOLV_CACHE_UNSUPPORTED;
+#endif
if (statp->nscount == 0) {
errno = ESRCH;
@@ -365,6 +377,20 @@
gotsomewhere = 0;
terrno = ETIMEDOUT;
+#if USE_RESOLV_CACHE
+ cache = __get_res_cache();
+ if (cache != NULL) {
+ int anslen = 0;
+ cache_status = _resolv_cache_lookup(
+ cache, buf, buflen,
+ ans, anssiz, &anslen);
+
+ if (cache_status == RESOLV_CACHE_FOUND) {
+ return anslen;
+ }
+ }
+#endif
+
/*
* If the ns_addr_list in the resolver context has changed, then
* invalidate our cached copy and the associated timing data.
@@ -534,6 +560,12 @@
(stdout, "%s", ""),
ans, (resplen > anssiz) ? anssiz : resplen);
+#if USE_RESOLV_CACHE
+ if (cache_status == RESOLV_CACHE_NOTFOUND) {
+ _resolv_cache_add(cache, buf, buflen,
+ ans, resplen);
+ }
+#endif
/*
* If we have temporarily opened a virtual circuit,
* or if we haven't been asked to keep a socket open,
diff --git a/libc/netbsd/resolv/res_state.c b/libc/netbsd/resolv/res_state.c
index 8f2851a..3a2301d 100644
--- a/libc/netbsd/resolv/res_state.c
+++ b/libc/netbsd/resolv/res_state.c
@@ -46,7 +46,6 @@
struct __res_state _nres[1];
unsigned _serial;
struct prop_info* _pi;
- struct hostent* _hostent;
struct res_static _rstatic[1];
} _res_thread;
@@ -66,9 +65,9 @@
if ( res_ninit( rt->_nres ) < 0 ) {
free(rt);
rt = NULL;
+ } else {
+ memset(rt->_rstatic, 0, sizeof rt->_rstatic);
}
- rt->_hostent = NULL;
- memset(rt->_rstatic, 0, sizeof rt->_rstatic);
}
return rt;
}
@@ -93,7 +92,6 @@
_res_thread* rt = _rt;
_res_static_done(rt->_rstatic);
- _resolv_hostent_free(rt->_hostent);
res_ndestroy(rt->_nres);
free(rt);
}
@@ -132,6 +130,7 @@
rt = NULL;
pthread_setspecific( _res_key, rt );
}
+ _resolv_cache_reset(rt->_serial);
return rt;
}
@@ -177,14 +176,6 @@
res=res;
}
-struct hostent**
-__get_res_cache_hostent_p(void)
-{
- _res_thread* rt = _res_thread_get();
-
- return rt ? &rt->_hostent : NULL;
-}
-
res_static
__res_get_static(void)
{
diff --git a/libc/bionic/logd.h b/libc/private/logd.h
similarity index 100%
rename from libc/bionic/logd.h
rename to libc/private/logd.h
diff --git a/libc/private/resolv_cache.h b/libc/private/resolv_cache.h
index 8c25583..cd876fb 100644
--- a/libc/private/resolv_cache.h
+++ b/libc/private/resolv_cache.h
@@ -28,31 +28,39 @@
#ifndef _RESOLV_CACHE_H_
#define _RESOLV_CACHE_H_
-#include <netdb.h>
-
-const struct hostent _resolv_hostent_none;
-#define _RESOLV_HOSTENT_NONE ((struct hostent*)&_resolv_hostent_none)
-
struct resolv_cache; /* forward */
+
+/* get cache instance, can be NULL if cache is disabled
+ * (e.g. through an environment variable) */
extern struct resolv_cache* __get_res_cache(void);
-extern struct hostent** __get_res_cache_hostent_p(void);
-extern struct resolv_cache* _resolv_cache_get( void );
+/* this gets called everytime we detect some changes in the DNS configuration
+ * and will flush the cache */
+extern void _resolv_cache_reset( unsigned generation );
-extern struct resolv_cache* _resolv_cache_create( void );
+typedef enum {
+ RESOLV_CACHE_UNSUPPORTED, /* the cache can't handle that kind of queries */
+ /* or the answer buffer is too small */
+ RESOLV_CACHE_NOTFOUND, /* the cache doesn't know about this query */
+ RESOLV_CACHE_FOUND /* the cache found the answer */
+} ResolvCacheStatus;
-extern void _resolv_cache_destroy( struct resolv_cache* cache );
+extern ResolvCacheStatus
+_resolv_cache_lookup( struct resolv_cache* cache,
+ const void* query,
+ int querylen,
+ void* answer,
+ int answersize,
+ int *answerlen );
-extern struct hostent* _resolv_cache_lookup( struct resolv_cache* cache,
- const char* name,
- int af );
-
-extern void _resolv_cache_add( struct resolv_cache* cache,
- const char* name,
- int af,
- struct hostent* hp );
-
-extern struct hostent* _resolv_hostent_copy( struct hostent* hp );
-extern void _resolv_hostent_free( struct hostent* hp );
+/* add a (query,answer) to the cache, only call if _resolv_cache_lookup
+ * did return RESOLV_CACHE_NOTFOUND
+ */
+extern void
+_resolv_cache_add( struct resolv_cache* cache,
+ const void* query,
+ int querylen,
+ const void* answer,
+ int answerlen );
#endif /* _RESOLV_CACHE_H_ */