sn1_brand.c revision 8fd04b8338ed5093ec2d1e668fa620b7de44c177
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright 2010 Sun Microsystems, Inc. All rights reserved.
* Use is subject to license terms.
*/
#include <errno.h>
#include <stdio.h>
#include <stdlib.h>
#include <strings.h>
#include <unistd.h>
#include <sys/inttypes.h>
#include <sn1_brand.h>
#include <sn1_misc.h>
/*
* Principles of emulation 101.
*
*
* *** Setting errno
*
* Just don't do it. This emulation library is loaded onto a
* seperate link map from the application who's address space we're
* running in. We have our own private copy of libc, so there for,
* the errno value accessible from here is is also private and changing
* it will not affect any errno value that the processes who's address
* space we are running in will see. To return an error condition we
* should return the negated errno value we'd like the system to return.
* For more information about this see the comment in sn1_handler().
* Basically, when we return to the caller that initiated the system
* call it's their responsibility to set errno.
*
*
* *** Recursion Considerations
*
* When emulating system calls we need to be very careful about what
* library calls we invoke. Library calls should be kept to a minimum.
* One issue is that library calls can invoke system calls, so if we're
* emulating a system call and we invoke a library call that depends on
* that system call we will probably enter a recursive loop, which would
* be bad.
*
*
* *** Return Values.
*
* When declaring new syscall emulation functions, it is very important
* to to set the proper RV_* flags in the sn1_sysent_table. Upon failure,
* syscall emulation fuctions should return an errno value. Upon success
* syscall emulation functions should return 0 and set the sysret_t return
* value parameters accordingly.
*
*
* *** Agent lwp considerations
*
* It is currently impossible to do any emulation for these system call
* when they are being invoked on behalf of an agent lwp. To understand why
* it's impossible you have to understand how agent lwp syscalls work.
*
* The agent lwp syscall process works as follows:
* 1 The controlling process stops the target.
* 2 The controlling process injects an agent lwp which is also stopped.
* This agent lwp assumes the userland stack and register values
* of another stopped lwp in the current process.
* 3 The controlling process configures the agent lwp to start
* executing the requested system call.
* 4 The controlling process configure /proc to stop the agent lwp when
* it enters the requested system call.
* 5 The controlling processes allows the agent lwp to start executing.
* 6 The agent lwp traps into the kernel to perform the requested system
* call and immediately stop.
* 7 The controlling process copies all the arguments for the requested
* system call onto the agent lwp's stack.
* 8 The controlling process configures /proc to stop the agent lwp
* when it completes the requested system call.
* 9 The controlling processes allows the agent lwp to start executing.
* 10 The agent lwp executes the system call and then stop before returning
* to userland.
* 11 The controlling process copies the return value and return arguments
* back from the agent lwps stack.
* 12 The controlling process destroys the agent lwp and restarts
* the target process.
*
* The fundamental problem is that when the agent executes the request
* system call in step 5, if we're emulating that system call then the
* lwp is redirected back to our emulation layer without blocking
* in the kernel. But our emulation layer can't access the arguments
* for the system call because they haven't been copied to the stack
* yet and they still only exist in the controlling processes address
* space. This prevents us from being able to do any emulation of
* agent lwp system calls. Hence, currently our brand trap interposition
* callback (sn1_brand_syscall_callback_common) will detect if a system
* call is being made by an agent lwp, and if this is the case it will
* never redirect the system call to this emulation library.
*
* In the future, if this proves to be a problem the the easiest solution
* would probably be to replace the branded versions of these application
* with their native counterparts. Ie, truss, plimit, and pfiles could be
* replace with wrapper scripts that execute the native versions of these
* applications. In the case of plimit and pfiles this should be pretty
* strait forward. Truss would probably be more tricky since it can
* execute applications which would be branded applications, so in that
* case it might be necessary to create a loadable library which could
* be LD_PRELOADed into truss and this library would interpose on the
* exec() system call to allow truss to correctly execute branded
* processes. It should be pointed out that this solution could work
* because "native agent lwps" (ie, agent lwps created by native
* processes) can be treated differently from "branded aged lwps" (ie,
* agent lwps created by branded processes), since native agent lwps
* would presumably be making native system calls and hence not need
* any interposition.
*
*
* *** sn1 brand emulation scope considerations
*
* One of the differences between the lx brand and the s8 and s9
* brands, is that the s8 and s9 brands only interpose on syscalls
* that need some kind of emulation, where as the lx brand interposes
* on _all_ system calls. Lx branded system calls that don't need
* any emulation are then redirected back to the kernel from the
* userland library via the IN_KERNEL_SYSCALL macro. The lx-syscall
* dtrace provider depends on this behavior.
*
* Given that the sn1 brand exists for testing purposes, it should
* eventually be enhanced to redirect all system calls through the
* brand emulation library. This will ensure the maximum testing
* exposure for the brandz infrastructure. Some other options to
* consider for improving brandz test exposure are:
* - Folding the sn1 brand into the native brand and only enabling
* it on DEBUG builds.
* - Modifying the zones test suite to use sn1 branded zones by default,
* any adapting functional test harnesses to use sn1 branded zones
* by default instead of native zones.
*/
typedef long (*sysent_cb_t)();
typedef struct sn1_sysent_table {
/*LINTED: static unused*/
static volatile int sn1_abort_err;
/*LINTED: static unused*/
static volatile const char *sn1_abort_msg;
/*LINTED: static unused*/
static volatile const char *sn1_abort_file;
/*LINTED: static unused*/
static volatile int sn1_abort_line;
extern int errno;
/*ARGSUSED*/
void
{
/* Save the error message into convenient globals */
sn1_abort_err = err;
sn1_abort_msg = msg;
/* kill ourselves */
abort();
/* If abort() didn't work, try something stronger. */
}
/*
* This function is defined to be NOSYS but it won't be called from the
* the kernel since the NOSYS system calls are not enabled in the kernel.
* Thus, the only time this function is called is directly from within the
* indirect system call path.
*/
/*ARGSUSED*/
static long
{
/*
* We'd like to print out some kind of error message here like
* "unsupported syscall", but we can't because it's not safe to
* assume that stderr or STDERR_FILENO actually points to something
* that is a terminal, and if we wrote to those files we could
* inadvertantly write to some applications open files, which would
* be bad.
*
* Normally, if an application calls an invalid system call
* it get a SIGSYS sent to it. So we'll just go ahead and send
* ourselves a signal here. Note that this is far from ideal since
* if the application has registered a signal handler, that signal
* handler may recieve a ucontext_t as the third parameter to
* indicate the context of the process when the signal was
* generated, and in this case that context will not be what the
* application is expecting. Hence, we should probably create a
* brandsys() kernel function that can deliver the signal to us
* with the correct ucontext_t.
*/
return (ENOSYS);
}
/*
* Yuck. For 32-bit sparc applications, handle indirect system calls.
* Note that we declare this interface to use the maximum number of
* system call arguments. If we recieve a system call that uses less
* arguments, then the additional arguments will be garbage, but they
* will also be ignored so that should be ok.
*/
static long
{
case 0:
case 1:
case 2:
case 3:
case 4:
case 5:
case 6:
case 7:
case 8:
}
sn1_abort(0, "invalid entry in sn1_sysent_table");
return (EINVAL);
}
#endif /* __sparc && !__sparcv9 */
static long
{
return (err);
return (EFAULT);
return (0);
}
/*
* Close a libc file handle, but don't actually close the underlying
* file descriptor.
*/
static void
{
return;
return;
if (fd_new == -1)
return;
}
/*ARGSUSED*/
int
{
uintptr_t *p;
int i, err;
/* Sanity check our translation table return value codes */
for (i = 0; i < NSYSCALL; i++) {
}
/*
* We need to shutdown all libc stdio. libc stdio normally goes to
* file descriptors, but since we're actually part of a another
* process we don't own these file descriptors and we can't make
* any assumptions about their state.
*/
/*
* Register our syscall emulation table with the kernel.
* Note that we don't have to do invoke (syscall_number + 1024)
* until we've actually establised a syscall emulation callback
* handler address, which is what we're doing with this brand
* syscall.
*/
#ifdef __x86
#else /* !__x86 */
#endif /* !__x86 */
/*NOTREACHED*/
}
/* Get data about the executable we're running from the kernel. */
"Failed to get required brand ELF data from the kernel");
/*NOTREACHED*/
}
/*
* Find the aux vector on the stack.
*/
while (*p != NULL)
p++;
/*
* p is now pointing at the 0 word after the environ pointers.
* After that is the aux vectors.
*
* The aux vectors are currently pointing to the brand emulation
* library and associated linker. We're going to change them to
* point to the brand executable and associated linker (or to no
* linker for static binaries). This matches the process data
* stored within the kernel and visible from /proc, which was
* all setup in sn1_elfexec(). We do this so that when a debugger
* attaches to the process it sees the process as a normal solaris
* process, this brand emulation library and everything on it's
* link map will not be visible, unless our librtld_db plugin
* is used. Note that this is very different from how Linux
* branded processes are implemented within lx branded zones.
* In that situation, the primary linkmap of the process is the
* brand emulation libraries linkmap, not the Linux applications
* linkmap.
*
* We also need to clear the AF_SUN_NOPLM flag from the AT_SUN_AUXFLAGS
* aux vector. This flag told our linker that we don't have a
* primary link map. Now that our linker is done initializing, we
* want to clear this flag before we transfer control to the
* applications copy of the linker, since we want that linker to have
* a primary link map which will be the link map for the application
* we're running.
*/
p++;
case AT_BASE:
/* Hide AT_BASE if static binary */
} else {
}
break;
case AT_ENTRY:
break;
case AT_PHDR:
break;
case AT_PHENT:
break;
case AT_PHNUM:
break;
case AT_SUN_AUXFLAGS:
break;
case AT_SUN_EMULATOR:
/*
* ld.so.1 inspects AT_SUN_EMULATOR to see if
* if it is the linker for the brand emulation
* library. Hide AT_SUN_EMULATOR, as the
* linker we are about to jump to is the linker
* for the binary.
*/
break;
case AT_SUN_LDDATA:
/* Hide AT_SUN_LDDATA if static binary */
} else {
}
break;
default:
break;
}
}
/*NOTREACHED*/
sn1_abort(0, "sn1_runexe() returned");
return (-1);
}
static long \
{ \
}
/*
* These are branded system calls, which have been redirected to this
* userland emulation library, and are emulated by passing them strait
* on to the kernel as native system calls.
*/
/*
* This table must have at least NSYSCALL entries in it.
*
* The second parameter of each entry in the sn1_sysent_table
* contains the number of parameters and flags that describe the
* syscall return value encoding. See the block comments at the
* top of this file for more information about the syscall return
* value flags and when they should be used.
*/
#else /* !__sparc || __sparcv9 */
NOSYS, /* 0 */
#endif /* !__sparc || __sparcv9 */
NOSYS, /* 1 */
NOSYS, /* 2 */
NOSYS, /* 5 */
NOSYS, /* 6 */
NOSYS, /* 7 */
NOSYS, /* 8 */
NOSYS, /* 9 */
NOSYS, /* 10 */
NOSYS, /* 11 */
NOSYS, /* 12 */
NOSYS, /* 14 */
NOSYS, /* 15 */
NOSYS, /* 16 */
NOSYS, /* 17 */
NOSYS, /* 18 */
NOSYS, /* 19 */
NOSYS, /* 22 */
NOSYS, /* 23 */
NOSYS, /* 25 */
NOSYS, /* 26 */
NOSYS, /* 27 */
NOSYS, /* 28 */
NOSYS, /* 29 */
NOSYS, /* 30 */
NOSYS, /* 31 */
NOSYS, /* 32 */
NOSYS, /* 33 */
NOSYS, /* 34 */
NOSYS, /* 35 */
NOSYS, /* 36 */
NOSYS, /* 37 */
NOSYS, /* 38 */
NOSYS, /* 39 */
NOSYS, /* 40 */
NOSYS, /* 41 */
NOSYS, /* 42 */
NOSYS, /* 44 */
NOSYS, /* 45 */
NOSYS, /* 46 */
NOSYS, /* 48 */
NOSYS, /* 49 */
NOSYS, /* 50 */
NOSYS, /* 51 */
NOSYS, /* 52 */
NOSYS, /* 53 */
NOSYS, /* 54 */
NOSYS, /* 55 */
NOSYS, /* 56 */
NOSYS, /* 58 */
NOSYS, /* 59 */
NOSYS, /* 60 */
NOSYS, /* 61 */
NOSYS, /* 62 */
NOSYS, /* 63 */
NOSYS, /* 64 */
NOSYS, /* 65 */
NOSYS, /* 66 */
NOSYS, /* 67 */
NOSYS, /* 68 */
NOSYS, /* 69 */
NOSYS, /* 70 */
NOSYS, /* 71 */
NOSYS, /* 72 */
NOSYS, /* 73 */
NOSYS, /* 74 */
NOSYS, /* 75 */
NOSYS, /* 76 */
NOSYS, /* 77 */
NOSYS, /* 78 */
NOSYS, /* 79 */
NOSYS, /* 80 */
NOSYS, /* 81 */
NOSYS, /* 82 */
NOSYS, /* 83 */
NOSYS, /* 84 */
NOSYS, /* 85 */
NOSYS, /* 86 */
NOSYS, /* 87 */
NOSYS, /* 88 */
NOSYS, /* 89 */
NOSYS, /* 91 */
NOSYS, /* 92 */
NOSYS, /* 93 */
NOSYS, /* 94 */
NOSYS, /* 95 */
NOSYS, /* 96 */
NOSYS, /* 97 */
NOSYS, /* 98 */
NOSYS, /* 99 */
NOSYS, /* 100 */
NOSYS, /* 101 */
NOSYS, /* 102 */
NOSYS, /* 103 */
NOSYS, /* 104 */
NOSYS, /* 105 */
NOSYS, /* 106 */
NOSYS, /* 108 */
NOSYS, /* 109 */
NOSYS, /* 110 */
NOSYS, /* 111 */
NOSYS, /* 112 */
NOSYS, /* 113 */
NOSYS, /* 114 */
NOSYS, /* 115 */
NOSYS, /* 116 */
NOSYS, /* 117 */
NOSYS, /* 118 */
NOSYS, /* 119 */
NOSYS, /* 120 */
NOSYS, /* 121 */
NOSYS, /* 122 */
NOSYS, /* 123 */
NOSYS, /* 124 */
NOSYS, /* 125 */
NOSYS, /* 126 */
NOSYS, /* 127 */
NOSYS, /* 128 */
NOSYS, /* 129 */
NOSYS, /* 130 */
NOSYS, /* 131 */
NOSYS, /* 132 */
NOSYS, /* 133 */
NOSYS, /* 134 */
NOSYS, /* 136 */
NOSYS, /* 137 */
NOSYS, /* 138 */
NOSYS, /* 139 */
NOSYS, /* 140 */
NOSYS, /* 141 */
NOSYS, /* 142 */
NOSYS, /* 143 */
NOSYS, /* 144 */
NOSYS, /* 145 */
NOSYS, /* 146 */
NOSYS, /* 147 */
NOSYS, /* 148 */
NOSYS, /* 149 */
NOSYS, /* 150 */
NOSYS, /* 151 */
NOSYS, /* 152 */
NOSYS, /* 153 */
NOSYS, /* 154 */
NOSYS, /* 155 */
NOSYS, /* 156 */
NOSYS, /* 157 */
NOSYS, /* 158 */
NOSYS, /* 159 */
NOSYS, /* 160 */
NOSYS, /* 161 */
NOSYS, /* 162 */
NOSYS, /* 163 */
NOSYS, /* 164 */
NOSYS, /* 165 */
NOSYS, /* 166 */
NOSYS, /* 167 */
NOSYS, /* 168 */
NOSYS, /* 169 */
NOSYS, /* 170 */
NOSYS, /* 171 */
NOSYS, /* 172 */
NOSYS, /* 173 */
NOSYS, /* 174 */
NOSYS, /* 175 */
NOSYS, /* 176 */
NOSYS, /* 177 */
NOSYS, /* 178 */
NOSYS, /* 179 */
NOSYS, /* 180 */
NOSYS, /* 181 */
NOSYS, /* 182 */
NOSYS, /* 183 */
NOSYS, /* 184 */
NOSYS, /* 185 */
NOSYS, /* 186 */
NOSYS, /* 187 */
NOSYS, /* 188 */
NOSYS, /* 189 */
NOSYS, /* 190 */
NOSYS, /* 191 */
NOSYS, /* 192 */
NOSYS, /* 193 */
NOSYS, /* 194 */
NOSYS, /* 195 */
NOSYS, /* 196 */
NOSYS, /* 197 */
NOSYS, /* 198 */
NOSYS, /* 199 */
NOSYS, /* 200 */
NOSYS, /* 201 */
NOSYS, /* 202 */
NOSYS, /* 203 */
NOSYS, /* 204 */
NOSYS, /* 205 */
NOSYS, /* 206 */
NOSYS, /* 207 */
NOSYS, /* 208 */
NOSYS, /* 209 */
NOSYS, /* 210 */
NOSYS, /* 211 */
NOSYS, /* 212 */
NOSYS, /* 213 */
NOSYS, /* 214 */
NOSYS, /* 215 */
NOSYS, /* 216 */
NOSYS, /* 217 */
NOSYS, /* 218 */
NOSYS, /* 219 */
NOSYS, /* 220 */
NOSYS, /* 221 */
NOSYS, /* 222 */
NOSYS, /* 223 */
NOSYS, /* 224 */
NOSYS, /* 225 */
NOSYS, /* 226 */
NOSYS, /* 227 */
NOSYS, /* 228 */
NOSYS, /* 229 */
NOSYS, /* 230 */
NOSYS, /* 231 */
NOSYS, /* 232 */
NOSYS, /* 233 */
NOSYS, /* 234 */
NOSYS, /* 235 */
NOSYS, /* 236 */
NOSYS, /* 237 */
NOSYS, /* 238 */
NOSYS, /* 239 */
NOSYS, /* 240 */
NOSYS, /* 241 */
NOSYS, /* 242 */
NOSYS, /* 243 */
NOSYS, /* 244 */
NOSYS, /* 245 */
NOSYS, /* 246 */
NOSYS, /* 247 */
NOSYS, /* 248 */
NOSYS, /* 249 */
NOSYS, /* 250 */
NOSYS, /* 251 */
NOSYS, /* 252 */
NOSYS, /* 253 */
NOSYS, /* 254 */
NOSYS /* 255 */
};