minix/servers/rs/exec.c

421 lines
12 KiB
C
Raw Normal View History

#include "inc.h"
#include <a.out.h>
#define BLOCK_SIZE 1024
static int do_exec(int proc_e, char *exec, size_t exec_len, char *progname,
char *frame, int frame_len);
FORWARD _PROTOTYPE( int read_header, (char *exec, size_t exec_len, int *sep_id,
vir_bytes *text_bytes, vir_bytes *data_bytes,
vir_bytes *bss_bytes, phys_bytes *tot_bytes, vir_bytes *pc,
int *hdrlenp) );
FORWARD _PROTOTYPE( int exec_newmem, (int proc_e, vir_bytes text_bytes,
vir_bytes data_bytes, vir_bytes bss_bytes, vir_bytes tot_bytes,
vir_bytes frame_len, int sep_id,
dev_t st_dev, ino_t st_ino, time_t st_ctime, char *progname,
int new_uid, int new_gid,
vir_bytes *stack_topp, int *load_textp, int *allow_setuidp) );
FORWARD _PROTOTYPE( int exec_restart, (int proc_e, int result,
vir_bytes pc) );
FORWARD _PROTOTYPE( void patch_ptr, (char stack[ARG_MAX],
vir_bytes base) );
FORWARD _PROTOTYPE( int read_seg, (char *exec, size_t exec_len, off_t off,
int proc_e, int seg, phys_bytes seg_bytes) );
New RS and new signal handling for system processes. UPDATING INFO: 20100317: /usr/src/etc/system.conf updated to ignore default kernel calls: copy it (or merge it) to /etc/system.conf. The hello driver (/dev/hello) added to the distribution: # cd /usr/src/commands/scripts && make clean install # cd /dev && MAKEDEV hello KERNEL CHANGES: - Generic signal handling support. The kernel no longer assumes PM as a signal manager for every process. The signal manager of a given process can now be specified in its privilege slot. When a signal has to be delivered, the kernel performs the lookup and forwards the signal to the appropriate signal manager. PM is the default signal manager for user processes, RS is the default signal manager for system processes. To enable ptrace()ing for system processes, it is sufficient to change the default signal manager to PM. This will temporarily disable crash recovery, though. - sys_exit() is now split into sys_exit() (i.e. exit() for system processes, which generates a self-termination signal), and sys_clear() (i.e. used by PM to ask the kernel to clear a process slot when a process exits). - Added a new kernel call (i.e. sys_update()) to swap two process slots and implement live update. PM CHANGES: - Posix signal handling is no longer allowed for system processes. System signals are split into two fixed categories: termination and non-termination signals. When a non-termination signaled is processed, PM transforms the signal into an IPC message and delivers the message to the system process. When a termination signal is processed, PM terminates the process. - PM no longer assumes itself as the signal manager for system processes. It now makes sure that every system signal goes through the kernel before being actually processes. The kernel will then dispatch the signal to the appropriate signal manager which may or may not be PM. SYSLIB CHANGES: - Simplified SEF init and LU callbacks. - Added additional predefined SEF callbacks to debug crash recovery and live update. - Fixed a temporary ack in the SEF init protocol. SEF init reply is now completely synchronous. - Added SEF signal event type to provide a uniform interface for system processes to deal with signals. A sef_cb_signal_handler() callback is available for system processes to handle every received signal. A sef_cb_signal_manager() callback is used by signal managers to process system signals on behalf of the kernel. - Fixed a few bugs with memory mapping and DS. VM CHANGES: - Page faults and memory requests coming from the kernel are now implemented using signals. - Added a new VM call to swap two process slots and implement live update. - The call is used by RS at update time and in turn invokes the kernel call sys_update(). RS CHANGES: - RS has been reworked with a better functional decomposition. - Better kernel call masks. com.h now defines the set of very basic kernel calls every system service is allowed to use. This makes system.conf simpler and easier to maintain. In addition, this guarantees a higher level of isolation for system libraries that use one or more kernel calls internally (e.g. printf). - RS is the default signal manager for system processes. By default, RS intercepts every signal delivered to every system process. This makes crash recovery possible before bringing PM and friends in the loop. - RS now supports fast rollback when something goes wrong while initializing the new version during a live update. - Live update is now implemented by keeping the two versions side-by-side and swapping the process slots when the old version is ready to update. - Crash recovery is now implemented by keeping the two versions side-by-side and cleaning up the old version only when the recovery process is complete. DS CHANGES: - Fixed a bug when the process doing ds_publish() or ds_delete() is not known by DS. - Fixed the completely broken support for strings. String publishing is now implemented in the system library and simply wraps publishing of memory ranges. Ideally, we should adopt a similar approach for other data types as well. - Test suite fixed. DRIVER CHANGES: - The hello driver has been added to the Minix distribution to demonstrate basic live update and crash recovery functionalities. - Other drivers have been adapted to conform the new SEF interface.
2010-03-17 02:15:29 +01:00
int srv_execve(int proc_e, char *exec, size_t exec_len, char **argv,
char **Xenvp)
{
char * const *ap;
char * const *ep;
char *frame;
char **vp;
char *sp, *progname;
size_t argc;
size_t frame_size;
size_t string_off;
size_t n;
int ov;
int r;
/* Assumptions: size_t and char *, it's all the same thing. */
/* Create a stack image that only needs to be patched up slightly
* by the kernel to be used for the process to be executed.
*/
ov= 0; /* No overflow yet. */
frame_size= 0; /* Size of the new initial stack. */
string_off= 0; /* Offset to start of the strings. */
argc= 0; /* Argument count. */
for (ap= argv; *ap != NULL; ap++) {
n = sizeof(*ap) + strlen(*ap) + 1;
frame_size+= n;
if (frame_size < n) ov= 1;
string_off+= sizeof(*ap);
argc++;
}
/* Add an argument count and two terminating nulls. */
frame_size+= sizeof(argc) + sizeof(*ap) + sizeof(*ep);
string_off+= sizeof(argc) + sizeof(*ap) + sizeof(*ep);
/* Align. */
frame_size= (frame_size + sizeof(char *) - 1) & ~(sizeof(char *) - 1);
/* The party is off if there is an overflow. */
if (ov || frame_size < 3 * sizeof(char *)) {
errno= E2BIG;
return -1;
}
/* Allocate space for the stack frame. */
frame = (char *) malloc(frame_size);
if (!frame) {
errno = E2BIG;
return -1;
}
/* Set arg count, init pointers to vector and string tables. */
* (size_t *) frame = argc;
vp = (char **) (frame + sizeof(argc));
sp = frame + string_off;
/* Load the argument vector and strings. */
for (ap= argv; *ap != NULL; ap++) {
*vp++= (char *) (sp - frame);
n= strlen(*ap) + 1;
memcpy(sp, *ap, n);
sp+= n;
}
*vp++= NULL;
#if 0
/* Load the environment vector and strings. */
for (ep= envp; *ep != NULL; ep++) {
*vp++= (char *) (sp - frame);
n= strlen(*ep) + 1;
memcpy(sp, *ep, n);
sp+= n;
}
#endif
*vp++= NULL;
/* Padding. */
while (sp < frame + frame_size) *sp++= 0;
(progname=strrchr(argv[0], '/')) ? progname++ : (progname=argv[0]);
r = do_exec(proc_e, exec, exec_len, progname, frame, frame_size);
/* Return the memory used for the frame and exit. */
free(frame);
return r;
}
static int do_exec(int proc_e, char *exec, size_t exec_len, char *progname,
char *frame, int frame_len)
{
int r;
int hdrlen, sep_id, load_text, allow_setuid;
int need_restart, error;
vir_bytes stack_top, vsp;
vir_bytes text_bytes, data_bytes, bss_bytes, pc;
phys_bytes tot_bytes;
off_t off;
uid_t new_uid;
gid_t new_gid;
need_restart= 0;
error= 0;
/* Read the file header and extract the segment sizes. */
r = read_header(exec, exec_len, &sep_id,
&text_bytes, &data_bytes, &bss_bytes,
&tot_bytes, &pc, &hdrlen);
if (r != OK)
{
printf("do_exec: read_header failed\n");
New RS and new signal handling for system processes. UPDATING INFO: 20100317: /usr/src/etc/system.conf updated to ignore default kernel calls: copy it (or merge it) to /etc/system.conf. The hello driver (/dev/hello) added to the distribution: # cd /usr/src/commands/scripts && make clean install # cd /dev && MAKEDEV hello KERNEL CHANGES: - Generic signal handling support. The kernel no longer assumes PM as a signal manager for every process. The signal manager of a given process can now be specified in its privilege slot. When a signal has to be delivered, the kernel performs the lookup and forwards the signal to the appropriate signal manager. PM is the default signal manager for user processes, RS is the default signal manager for system processes. To enable ptrace()ing for system processes, it is sufficient to change the default signal manager to PM. This will temporarily disable crash recovery, though. - sys_exit() is now split into sys_exit() (i.e. exit() for system processes, which generates a self-termination signal), and sys_clear() (i.e. used by PM to ask the kernel to clear a process slot when a process exits). - Added a new kernel call (i.e. sys_update()) to swap two process slots and implement live update. PM CHANGES: - Posix signal handling is no longer allowed for system processes. System signals are split into two fixed categories: termination and non-termination signals. When a non-termination signaled is processed, PM transforms the signal into an IPC message and delivers the message to the system process. When a termination signal is processed, PM terminates the process. - PM no longer assumes itself as the signal manager for system processes. It now makes sure that every system signal goes through the kernel before being actually processes. The kernel will then dispatch the signal to the appropriate signal manager which may or may not be PM. SYSLIB CHANGES: - Simplified SEF init and LU callbacks. - Added additional predefined SEF callbacks to debug crash recovery and live update. - Fixed a temporary ack in the SEF init protocol. SEF init reply is now completely synchronous. - Added SEF signal event type to provide a uniform interface for system processes to deal with signals. A sef_cb_signal_handler() callback is available for system processes to handle every received signal. A sef_cb_signal_manager() callback is used by signal managers to process system signals on behalf of the kernel. - Fixed a few bugs with memory mapping and DS. VM CHANGES: - Page faults and memory requests coming from the kernel are now implemented using signals. - Added a new VM call to swap two process slots and implement live update. - The call is used by RS at update time and in turn invokes the kernel call sys_update(). RS CHANGES: - RS has been reworked with a better functional decomposition. - Better kernel call masks. com.h now defines the set of very basic kernel calls every system service is allowed to use. This makes system.conf simpler and easier to maintain. In addition, this guarantees a higher level of isolation for system libraries that use one or more kernel calls internally (e.g. printf). - RS is the default signal manager for system processes. By default, RS intercepts every signal delivered to every system process. This makes crash recovery possible before bringing PM and friends in the loop. - RS now supports fast rollback when something goes wrong while initializing the new version during a live update. - Live update is now implemented by keeping the two versions side-by-side and swapping the process slots when the old version is ready to update. - Crash recovery is now implemented by keeping the two versions side-by-side and cleaning up the old version only when the recovery process is complete. DS CHANGES: - Fixed a bug when the process doing ds_publish() or ds_delete() is not known by DS. - Fixed the completely broken support for strings. String publishing is now implemented in the system library and simply wraps publishing of memory ranges. Ideally, we should adopt a similar approach for other data types as well. - Test suite fixed. DRIVER CHANGES: - The hello driver has been added to the Minix distribution to demonstrate basic live update and crash recovery functionalities. - Other drivers have been adapted to conform the new SEF interface.
2010-03-17 02:15:29 +01:00
error= r;
goto fail;
}
need_restart= 1;
new_uid= getuid();
new_gid= getgid();
/* XXX what should we use to identify the executable? */
r= exec_newmem(proc_e, text_bytes, data_bytes, bss_bytes, tot_bytes,
frame_len, sep_id, 0 /*dev*/, proc_e /*inum*/, 0 /*ctime*/,
progname, new_uid, new_gid, &stack_top, &load_text,
&allow_setuid);
if (r != OK)
{
printf("do_exec: exec_newmap failed: %d\n", r);
error= r;
goto fail;
}
/* Patch up stack and copy it from RS to new core image. */
vsp = stack_top;
vsp -= frame_len;
patch_ptr(frame, vsp);
r = sys_datacopy(SELF, (vir_bytes) frame,
proc_e, (vir_bytes) vsp, (phys_bytes)frame_len);
if (r != OK) {
printf("RS: stack_top is 0x%lx; tried to copy to 0x%lx in %d\n",
stack_top, vsp, proc_e);
printf("do_exec: copying out new stack failed: %d\n", r);
error= r;
goto fail;
}
off = hdrlen;
/* Read in text and data segments. */
if (load_text) {
r= read_seg(exec, exec_len, off, proc_e, T, text_bytes);
if (r != OK)
{
printf("do_exec: read_seg failed: %d\n", r);
error= r;
goto fail;
}
}
else
printf("do_exec: not loading text segment\n");
off += text_bytes;
r= read_seg(exec, exec_len, off, proc_e, D, data_bytes);
if (r != OK)
{
printf("do_exec: read_seg failed: %d\n", r);
error= r;
goto fail;
}
return exec_restart(proc_e, OK, pc);
fail:
printf("do_exec(fail): error = %d\n", error);
if (need_restart)
exec_restart(proc_e, error, pc);
return error;
}
/*===========================================================================*
* exec_newmem *
*===========================================================================*/
PRIVATE int exec_newmem(
int proc_e,
vir_bytes text_bytes,
vir_bytes data_bytes,
vir_bytes bss_bytes,
vir_bytes tot_bytes,
vir_bytes frame_len,
int sep_id,
dev_t st_dev,
ino_t st_ino,
time_t st_ctime,
char *progname,
int new_uid,
int new_gid,
vir_bytes *stack_topp,
int *load_textp,
int *allow_setuidp
)
{
int r;
struct exec_newmem e;
message m;
e.text_bytes= text_bytes;
e.data_bytes= data_bytes;
e.bss_bytes= bss_bytes;
e.tot_bytes= tot_bytes;
e.args_bytes= frame_len;
e.sep_id= sep_id;
e.st_dev= st_dev;
e.st_ino= st_ino;
e.st_ctime= st_ctime;
e.new_uid= new_uid;
e.new_gid= new_gid;
strncpy(e.progname, progname, sizeof(e.progname)-1);
e.progname[sizeof(e.progname)-1]= '\0';
m.m_type= EXEC_NEWMEM;
m.EXC_NM_PROC= proc_e;
m.EXC_NM_PTR= (char *)&e;
r= sendrec(PM_PROC_NR, &m);
if (r != OK)
return r;
#if 0
printf("exec_newmem: r = %d, m_type = %d\n", r, m.m_type);
#endif
*stack_topp= m.m1_i1;
*load_textp= !!(m.m1_i2 & EXC_NM_RF_LOAD_TEXT);
*allow_setuidp= !!(m.m1_i2 & EXC_NM_RF_ALLOW_SETUID);
#if 0
printf("RS: exec_newmem: stack_top = 0x%x\n", *stack_topp);
printf("RS: exec_newmem: load_text = %d\n", *load_textp);
#endif
return m.m_type;
}
/*===========================================================================*
* exec_restart *
*===========================================================================*/
PRIVATE int exec_restart(proc_e, result, pc)
int proc_e;
int result;
vir_bytes pc;
{
int r;
message m;
m.m_type= EXEC_RESTART;
m.EXC_RS_PROC= proc_e;
m.EXC_RS_RESULT= result;
m.EXC_RS_PC= (void*)pc;
r= sendrec(PM_PROC_NR, &m);
if (r != OK)
return r;
return m.m_type;
}
/*===========================================================================*
* read_header *
*===========================================================================*/
PRIVATE int read_header(exec, exec_len, sep_id, text_bytes, data_bytes,
bss_bytes, tot_bytes, pc, hdrlenp)
char *exec; /* executable image */
size_t exec_len; /* size of the image */
int *sep_id; /* true iff sep I&D */
vir_bytes *text_bytes; /* place to return text size */
vir_bytes *data_bytes; /* place to return initialized data size */
vir_bytes *bss_bytes; /* place to return bss size */
phys_bytes *tot_bytes; /* place to return total size */
vir_bytes *pc; /* program entry point (initial PC) */
int *hdrlenp;
{
/* Read the header and extract the text, data, bss and total sizes from it. */
struct exec hdr; /* a.out header is read in here */
/* Read the header and check the magic number. The standard MINIX header
* is defined in <a.out.h>. It consists of 8 chars followed by 6 longs.
* Then come 4 more longs that are not used here.
* Byte 0: magic number 0x01
* Byte 1: magic number 0x03
* Byte 2: normal = 0x10 (not checked, 0 is OK), separate I/D = 0x20
* Byte 3: CPU type, Intel 16 bit = 0x04, Intel 32 bit = 0x10,
* Motorola = 0x0B, Sun SPARC = 0x17
* Byte 4: Header length = 0x20
* Bytes 5-7 are not used.
*
* Now come the 6 longs
* Bytes 8-11: size of text segments in bytes
* Bytes 12-15: size of initialized data segment in bytes
* Bytes 16-19: size of bss in bytes
* Bytes 20-23: program entry point
* Bytes 24-27: total memory allocated to program (text, data + stack)
* Bytes 28-31: size of symbol table in bytes
* The longs are represented in a machine dependent order,
* little-endian on the 8088, big-endian on the 68000.
* The header is followed directly by the text and data segments, and the
* symbol table (if any). The sizes are given in the header. Only the
* text and data segments are copied into memory by exec. The header is
* used here only. The symbol table is for the benefit of a debugger and
* is ignored here.
*/
if (exec_len < sizeof(hdr)) return(ENOEXEC);
memcpy(&hdr, exec, sizeof(hdr));
/* Check magic number, cpu type, and flags. */
if (BADMAG(hdr)) return(ENOEXEC);
#if (CHIP == INTEL && _WORD_SIZE == 2)
if (hdr.a_cpu != A_I8086) return(ENOEXEC);
#endif
#if (CHIP == INTEL && _WORD_SIZE == 4)
if (hdr.a_cpu != A_I80386) return(ENOEXEC);
#endif
if ((hdr.a_flags & ~(A_NSYM | A_EXEC | A_SEP)) != 0) return(ENOEXEC);
*sep_id = !!(hdr.a_flags & A_SEP); /* separate I & D or not */
/* Get text and data sizes. */
*text_bytes = (vir_bytes) hdr.a_text; /* text size in bytes */
*data_bytes = (vir_bytes) hdr.a_data; /* data size in bytes */
*bss_bytes = (vir_bytes) hdr.a_bss; /* bss size in bytes */
*tot_bytes = hdr.a_total; /* total bytes to allocate for prog */
if (*tot_bytes == 0) return(ENOEXEC);
if (!*sep_id) {
/* If I & D space is not separated, it is all considered data. Text=0*/
*data_bytes += *text_bytes;
*text_bytes = 0;
}
*pc = hdr.a_entry; /* initial address to start execution */
*hdrlenp = hdr.a_hdrlen & BYTE; /* header length */
return(OK);
}
/*===========================================================================*
* patch_ptr *
*===========================================================================*/
PRIVATE void patch_ptr(stack, base)
char stack[ARG_MAX]; /* pointer to stack image within PM */
vir_bytes base; /* virtual address of stack base inside user */
{
/* When doing an exec(name, argv, envp) call, the user builds up a stack
* image with arg and env pointers relative to the start of the stack. Now
* these pointers must be relocated, since the stack is not positioned at
* address 0 in the user's address space.
*/
char **ap, flag;
vir_bytes v;
flag = 0; /* counts number of 0-pointers seen */
ap = (char **) stack; /* points initially to 'nargs' */
ap++; /* now points to argv[0] */
while (flag < 2) {
if (ap >= (char **) &stack[ARG_MAX]) return; /* too bad */
if (*ap != NULL) {
v = (vir_bytes) *ap; /* v is relative pointer */
v += base; /* relocate it */
*ap = (char *) v; /* put it back */
} else {
flag++;
}
ap++;
}
}
/*===========================================================================*
* read_seg *
*===========================================================================*/
PRIVATE int read_seg(exec, exec_len, off, proc_e, seg, seg_bytes)
char *exec; /* executable image */
size_t exec_len; /* size of the image */
off_t off; /* offset in file */
int proc_e; /* process number (endpoint) */
int seg; /* T, D, or S */
phys_bytes seg_bytes; /* how much is to be transferred? */
{
/*
* The byte count on read is usually smaller than the segment count, because
* a segment is padded out to a click multiple, and the data segment is only
* partially initialized.
*/
int r;
if (off+seg_bytes > exec_len) return ENOEXEC;
r= sys_vircopy(SELF, D, (vir_bytes)exec+off, proc_e, seg, 0, seg_bytes);
return r;
}