Index: head/sys/compat/linprocfs/linprocfs.c =================================================================== --- head/sys/compat/linprocfs/linprocfs.c (revision 60859) +++ head/sys/compat/linprocfs/linprocfs.c (revision 60860) @@ -1,202 +1,209 @@ /* * Copyright (c) 1993 Jan-Simon Pendry * Copyright (c) 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Jan-Simon Pendry. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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. * * @(#)procfs_status.c 8.4 (Berkeley) 6/15/94 * * $FreeBSD$ */ #include #include #include #include #include #include #include #include #include #include #include #include +#include #include #include #include #include #include struct proc; int linprocfs_domeminfo(curp, p, pfs, uio) struct proc *curp; struct proc *p; struct pfsnode *pfs; struct uio *uio; { char *ps; int xlen; int error; char psbuf[512]; /* XXX - conservative */ unsigned long memtotal; /* total memory in bytes */ unsigned long memused; /* used memory in bytes */ unsigned long memfree; /* free memory in bytes */ unsigned long memshared; /* shared memory ??? */ unsigned long buffers, cached; /* buffer / cache memory ??? */ unsigned long swaptotal; /* total swap space in bytes */ unsigned long swapused; /* used swap space in bytes */ unsigned long swapfree; /* free swap space in bytes */ + vm_object_t object; if (uio->uio_rw != UIO_READ) return (EOPNOTSUPP); memtotal = physmem * PAGE_SIZE; /* * The correct thing here would be: * memfree = cnt.v_free_count * PAGE_SIZE; memused = memtotal - memfree; * * but it might mislead linux binaries into thinking there * is very little memory left, so we cheat and tell them that * all memory that isn't wired down is free. */ memused = cnt.v_wire_count * PAGE_SIZE; memfree = memtotal - memused; swaptotal = swapblist->bl_blocks * 1024; /* XXX why 1024? */ swapfree = swapblist->bl_root->u.bmu_avail * PAGE_SIZE; swapused = swaptotal - swapfree; - memshared = 0; /* XXX what's this supposed to be? */ + memshared = 0; + for (object = TAILQ_FIRST(&vm_object_list); object != NULL; + object = TAILQ_NEXT(object, object_list)) + if (object->shadow_count > 1) + memshared += object->resident_page_count; + memshared *= PAGE_SIZE; /* * We'd love to be able to write: * buffers = bufspace; * * but bufspace is internal to vfs_bio.c and we don't feel * like unstaticizing it just for linprocfs's sake. */ buffers = 0; cached = cnt.v_cache_count * PAGE_SIZE; ps = psbuf; ps += sprintf(ps, " total: used: free: shared: buffers: cached:\n" "Mem: %lu %lu %lu %lu %lu %lu\n" "Swap: %lu %lu %lu\n" "MemTotal: %9lu kB\n" "MemFree: %9lu kB\n" "MemShared:%9lu kB\n" "Buffers: %9lu kB\n" "Cached: %9lu kB\n" "SwapTotal:%9lu kB\n" "SwapFree: %9lu kB\n", memtotal, memused, memfree, memshared, buffers, cached, swaptotal, swapused, swapfree, memtotal >> 10, memfree >> 10, memshared >> 10, buffers >> 10, cached >> 10, swaptotal >> 10, swapfree >> 10); xlen = ps - psbuf; xlen -= uio->uio_offset; ps = psbuf + uio->uio_offset; xlen = imin(xlen, uio->uio_resid); if (xlen <= 0) error = 0; else error = uiomove(ps, xlen, uio); return (error); } int linprocfs_docpuinfo(curp, p, pfs, uio) struct proc *curp; struct proc *p; struct pfsnode *pfs; struct uio *uio; { char *ps; int xlen; int error; char psbuf[512]; /* XXX - conservative */ char *class; #if 0 extern char *cpu_model; /* Yuck */ #endif if (uio->uio_rw != UIO_READ) return (EOPNOTSUPP); switch (cpu_class) { case CPUCLASS_286: class = "286"; break; case CPUCLASS_386: class = "386"; break; case CPUCLASS_486: class = "486"; break; case CPUCLASS_586: class = "586"; break; case CPUCLASS_686: class = "686"; break; default: class = "unknown"; break; } ps = psbuf; ps += sprintf(ps, "processor : %d\n" "cpu : %.3s\n" "model : %.20s\n" "vendor_id : %.20s\n" "stepping : %d\n", 0, class, "unknown", cpu_vendor, cpu_id); xlen = ps - psbuf; xlen -= uio->uio_offset; ps = psbuf + uio->uio_offset; xlen = imin(xlen, uio->uio_resid); if (xlen <= 0) error = 0; else error = uiomove(ps, xlen, uio); return (error); } Index: head/sys/compat/linprocfs/linprocfs.h =================================================================== --- head/sys/compat/linprocfs/linprocfs.h (revision 60859) +++ head/sys/compat/linprocfs/linprocfs.h (revision 60860) @@ -1,135 +1,134 @@ /* * Copyright (c) 1993 Jan-Simon Pendry * Copyright (c) 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Jan-Simon Pendry. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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. * * @(#)procfs.h 8.9 (Berkeley) 5/14/95 * * $FreeBSD$ */ /* * The different types of node in a procfs filesystem */ typedef enum { Proot, /* the filesystem root */ Pself, /* symbolic link for curproc */ Pproc, /* a process-specific sub-directory */ Pexe, /* the executable file */ Pmem, /* the process's memory image */ Pmeminfo, Pcpuinfo } pfstype; /* * control data for the proc file system. */ struct pfsnode { struct pfsnode *pfs_next; /* next on list */ struct vnode *pfs_vnode; /* vnode associated with this pfsnode */ pfstype pfs_type; /* type of procfs node */ pid_t pfs_pid; /* associated process */ u_short pfs_mode; /* mode bits for stat() */ u_long pfs_flags; /* open flags */ u_long pfs_fileno; /* unique file id */ pid_t pfs_lockowner; /* pfs lock owner */ }; #define PROCFS_NAMELEN 8 /* max length of a filename component */ /* * Kernel stuff follows */ #ifdef _KERNEL #define CNEQ(cnp, s, len) \ ((cnp)->cn_namelen == (len) && \ (bcmp((s), (cnp)->cn_nameptr, (len)) == 0)) #define KMEM_GROUP 2 #define PROCFS_FILENO(pid, type) \ (((type) < Pproc) ? \ ((type) + 2) : \ ((((pid)+1) << 4) + ((int) (type)))) /* * Convert between pfsnode vnode */ #define VTOPFS(vp) ((struct pfsnode *)(vp)->v_data) #define PFSTOV(pfs) ((pfs)->pfs_vnode) typedef struct vfs_namemap vfs_namemap_t; struct vfs_namemap { const char *nm_name; int nm_val; }; int vfs_getuserstr __P((struct uio *, char *, int *)); vfs_namemap_t *vfs_findname __P((vfs_namemap_t *, char *, int)); /* */ struct reg; struct fpreg; struct dbreg; #define PFIND(pid) ((pid) ? pfind(pid) : &proc0) void linprocfs_exit __P((struct proc *)); int linprocfs_freevp __P((struct vnode *)); int linprocfs_allocvp __P((struct mount *, struct vnode **, long, pfstype)); -struct vnode *linprocfs_findtextvp __P((struct proc *)); int linprocfs_sstep __P((struct proc *)); void linprocfs_fix_sstep __P((struct proc *)); #if 0 int linprocfs_read_regs __P((struct proc *, struct reg *)); int linprocfs_write_regs __P((struct proc *, struct reg *)); int linprocfs_read_fpregs __P((struct proc *, struct fpreg *)); int linprocfs_write_fpregs __P((struct proc *, struct fpreg *)); int linprocfs_read_dbregs __P((struct proc *, struct dbreg *)); int linprocfs_write_dbregs __P((struct proc *, struct dbreg *)); #endif int linprocfs_domeminfo __P((struct proc *, struct proc *, struct pfsnode *pfsp, struct uio *uio)); int linprocfs_docpuinfo __P((struct proc *, struct proc *, struct pfsnode *pfsp, struct uio *uio)); /* functions to check whether or not files should be displayed */ int linprocfs_validfile __P((struct proc *)); #define PROCFS_LOCKED 0x01 #define PROCFS_WANT 0x02 extern vop_t **linprocfs_vnodeop_p; int linprocfs_root __P((struct mount *, struct vnode **)); int linprocfs_rw __P((struct vop_read_args *)); #endif /* _KERNEL */ Index: head/sys/compat/linprocfs/linprocfs_misc.c =================================================================== --- head/sys/compat/linprocfs/linprocfs_misc.c (revision 60859) +++ head/sys/compat/linprocfs/linprocfs_misc.c (revision 60860) @@ -1,202 +1,209 @@ /* * Copyright (c) 1993 Jan-Simon Pendry * Copyright (c) 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Jan-Simon Pendry. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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. * * @(#)procfs_status.c 8.4 (Berkeley) 6/15/94 * * $FreeBSD$ */ #include #include #include #include #include #include #include #include #include #include #include #include +#include #include #include #include #include #include struct proc; int linprocfs_domeminfo(curp, p, pfs, uio) struct proc *curp; struct proc *p; struct pfsnode *pfs; struct uio *uio; { char *ps; int xlen; int error; char psbuf[512]; /* XXX - conservative */ unsigned long memtotal; /* total memory in bytes */ unsigned long memused; /* used memory in bytes */ unsigned long memfree; /* free memory in bytes */ unsigned long memshared; /* shared memory ??? */ unsigned long buffers, cached; /* buffer / cache memory ??? */ unsigned long swaptotal; /* total swap space in bytes */ unsigned long swapused; /* used swap space in bytes */ unsigned long swapfree; /* free swap space in bytes */ + vm_object_t object; if (uio->uio_rw != UIO_READ) return (EOPNOTSUPP); memtotal = physmem * PAGE_SIZE; /* * The correct thing here would be: * memfree = cnt.v_free_count * PAGE_SIZE; memused = memtotal - memfree; * * but it might mislead linux binaries into thinking there * is very little memory left, so we cheat and tell them that * all memory that isn't wired down is free. */ memused = cnt.v_wire_count * PAGE_SIZE; memfree = memtotal - memused; swaptotal = swapblist->bl_blocks * 1024; /* XXX why 1024? */ swapfree = swapblist->bl_root->u.bmu_avail * PAGE_SIZE; swapused = swaptotal - swapfree; - memshared = 0; /* XXX what's this supposed to be? */ + memshared = 0; + for (object = TAILQ_FIRST(&vm_object_list); object != NULL; + object = TAILQ_NEXT(object, object_list)) + if (object->shadow_count > 1) + memshared += object->resident_page_count; + memshared *= PAGE_SIZE; /* * We'd love to be able to write: * buffers = bufspace; * * but bufspace is internal to vfs_bio.c and we don't feel * like unstaticizing it just for linprocfs's sake. */ buffers = 0; cached = cnt.v_cache_count * PAGE_SIZE; ps = psbuf; ps += sprintf(ps, " total: used: free: shared: buffers: cached:\n" "Mem: %lu %lu %lu %lu %lu %lu\n" "Swap: %lu %lu %lu\n" "MemTotal: %9lu kB\n" "MemFree: %9lu kB\n" "MemShared:%9lu kB\n" "Buffers: %9lu kB\n" "Cached: %9lu kB\n" "SwapTotal:%9lu kB\n" "SwapFree: %9lu kB\n", memtotal, memused, memfree, memshared, buffers, cached, swaptotal, swapused, swapfree, memtotal >> 10, memfree >> 10, memshared >> 10, buffers >> 10, cached >> 10, swaptotal >> 10, swapfree >> 10); xlen = ps - psbuf; xlen -= uio->uio_offset; ps = psbuf + uio->uio_offset; xlen = imin(xlen, uio->uio_resid); if (xlen <= 0) error = 0; else error = uiomove(ps, xlen, uio); return (error); } int linprocfs_docpuinfo(curp, p, pfs, uio) struct proc *curp; struct proc *p; struct pfsnode *pfs; struct uio *uio; { char *ps; int xlen; int error; char psbuf[512]; /* XXX - conservative */ char *class; #if 0 extern char *cpu_model; /* Yuck */ #endif if (uio->uio_rw != UIO_READ) return (EOPNOTSUPP); switch (cpu_class) { case CPUCLASS_286: class = "286"; break; case CPUCLASS_386: class = "386"; break; case CPUCLASS_486: class = "486"; break; case CPUCLASS_586: class = "586"; break; case CPUCLASS_686: class = "686"; break; default: class = "unknown"; break; } ps = psbuf; ps += sprintf(ps, "processor : %d\n" "cpu : %.3s\n" "model : %.20s\n" "vendor_id : %.20s\n" "stepping : %d\n", 0, class, "unknown", cpu_vendor, cpu_id); xlen = ps - psbuf; xlen -= uio->uio_offset; ps = psbuf + uio->uio_offset; xlen = imin(xlen, uio->uio_resid); if (xlen <= 0) error = 0; else error = uiomove(ps, xlen, uio); return (error); } Index: head/sys/compat/linprocfs/linprocfs_subr.c =================================================================== --- head/sys/compat/linprocfs/linprocfs_subr.c (revision 60859) +++ head/sys/compat/linprocfs/linprocfs_subr.c (revision 60860) @@ -1,353 +1,359 @@ /* * Copyright (c) 1993 Jan-Simon Pendry * Copyright (c) 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Jan-Simon Pendry. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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. * * @(#)procfs_subr.c 8.6 (Berkeley) 5/14/95 * * $FreeBSD$ */ #include #include #include #include #include #include static struct pfsnode *pfshead; static int pfsvplock; extern int procfs_domem __P((struct proc *, struct proc *, struct pfsnode *pfsp, struct uio *uio)); /* * allocate a pfsnode/vnode pair. the vnode is * referenced, but not locked. * * the pid, pfs_type, and mount point uniquely * identify a pfsnode. the mount point is needed * because someone might mount this filesystem * twice. * * all pfsnodes are maintained on a singly-linked * list. new nodes are only allocated when they cannot * be found on this list. entries on the list are * removed when the vfs reclaim entry is called. * * a single lock is kept for the entire list. this is * needed because the getnewvnode() function can block * waiting for a vnode to become free, in which case there * may be more than one process trying to get the same * vnode. this lock is only taken if we are going to * call getnewvnode, since the kernel itself is single-threaded. * * if an entry is found on the list, then call vget() to * take a reference. this is done because there may be * zero references to it and so it needs to removed from * the vnode free list. */ int linprocfs_allocvp(mp, vpp, pid, pfs_type) struct mount *mp; struct vnode **vpp; long pid; pfstype pfs_type; { struct proc *p = curproc; /* XXX */ struct pfsnode *pfs; struct vnode *vp; struct pfsnode **pp; int error; loop: for (pfs = pfshead; pfs != 0; pfs = pfs->pfs_next) { vp = PFSTOV(pfs); if (pfs->pfs_pid == pid && pfs->pfs_type == pfs_type && vp->v_mount == mp) { if (vget(vp, 0, p)) goto loop; *vpp = vp; return (0); } } /* * otherwise lock the vp list while we call getnewvnode * since that can block. */ if (pfsvplock & PROCFS_LOCKED) { pfsvplock |= PROCFS_WANT; (void) tsleep((caddr_t) &pfsvplock, PINOD, "pfsavp", 0); goto loop; } pfsvplock |= PROCFS_LOCKED; /* * Do the MALLOC before the getnewvnode since doing so afterward * might cause a bogus v_data pointer to get dereferenced * elsewhere if MALLOC should block. */ MALLOC(pfs, struct pfsnode *, sizeof(struct pfsnode), M_TEMP, M_WAITOK); if ((error = getnewvnode(VT_PROCFS, mp, linprocfs_vnodeop_p, vpp)) != 0) { FREE(pfs, M_TEMP); goto out; } vp = *vpp; vp->v_data = pfs; pfs->pfs_next = 0; pfs->pfs_pid = (pid_t) pid; pfs->pfs_type = pfs_type; pfs->pfs_vnode = vp; pfs->pfs_flags = 0; pfs->pfs_lockowner = 0; pfs->pfs_fileno = PROCFS_FILENO(pid, pfs_type); switch (pfs_type) { case Proot: /* /proc = dr-xr-xr-x */ pfs->pfs_mode = (VREAD|VEXEC) | (VREAD|VEXEC) >> 3 | (VREAD|VEXEC) >> 6; vp->v_type = VDIR; vp->v_flag = VROOT; break; case Pself: /* /proc/self = lr--r--r-- */ pfs->pfs_mode = (VREAD) | (VREAD >> 3) | (VREAD >> 6); vp->v_type = VLNK; break; case Pproc: pfs->pfs_mode = (VREAD|VEXEC) | (VREAD|VEXEC) >> 3 | (VREAD|VEXEC) >> 6; vp->v_type = VDIR; break; case Pexe: + pfs->pfs_mode = (VREAD|VEXEC) | + (VREAD|VEXEC) >> 3 | + (VREAD|VEXEC) >> 6; + vp->v_type = VLNK; + break; + case Pmem: pfs->pfs_mode = (VREAD|VWRITE) | (VREAD) >> 3;; vp->v_type = VREG; break; case Pmeminfo: case Pcpuinfo: pfs->pfs_mode = (VREAD) | (VREAD >> 3) | (VREAD >> 6); vp->v_type = VREG; break; default: panic("linprocfs_allocvp"); } /* add to procfs vnode list */ for (pp = &pfshead; *pp; pp = &(*pp)->pfs_next) continue; *pp = pfs; out: pfsvplock &= ~PROCFS_LOCKED; if (pfsvplock & PROCFS_WANT) { pfsvplock &= ~PROCFS_WANT; wakeup((caddr_t) &pfsvplock); } return (error); } int linprocfs_freevp(vp) struct vnode *vp; { struct pfsnode **pfspp; struct pfsnode *pfs = VTOPFS(vp); for (pfspp = &pfshead; *pfspp != 0; pfspp = &(*pfspp)->pfs_next) { if (*pfspp == pfs) { *pfspp = pfs->pfs_next; break; } } FREE(vp->v_data, M_TEMP); vp->v_data = 0; return (0); } int linprocfs_rw(ap) struct vop_read_args *ap; { struct vnode *vp = ap->a_vp; struct uio *uio = ap->a_uio; struct proc *curp = uio->uio_procp; struct pfsnode *pfs = VTOPFS(vp); struct proc *p; int rtval; p = PFIND(pfs->pfs_pid); if (p == 0) return (EINVAL); if (p->p_pid == 1 && securelevel > 0 && uio->uio_rw == UIO_WRITE) return (EACCES); while (pfs->pfs_lockowner) { tsleep(&pfs->pfs_lockowner, PRIBIO, "pfslck", 0); } pfs->pfs_lockowner = curproc->p_pid; switch (pfs->pfs_type) { case Pmem: rtval = procfs_domem(curp, p, pfs, uio); break; case Pmeminfo: rtval = linprocfs_domeminfo(curp, p, pfs, uio); break; case Pcpuinfo: rtval = linprocfs_docpuinfo(curp, p, pfs, uio); break; default: rtval = EOPNOTSUPP; break; } pfs->pfs_lockowner = 0; wakeup(&pfs->pfs_lockowner); return rtval; } #if 0 /* * Get a string from userland into (buf). Strip a trailing * nl character (to allow easy access from the shell). * The buffer should be *buflenp + 1 chars long. vfs_getuserstr * will automatically add a nul char at the end. * * Returns 0 on success or the following errors * * EINVAL: file offset is non-zero. * EMSGSIZE: message is longer than kernel buffer * EFAULT: user i/o buffer is not addressable */ int vfs_getuserstr(uio, buf, buflenp) struct uio *uio; char *buf; int *buflenp; { int xlen; int error; if (uio->uio_offset != 0) return (EINVAL); xlen = *buflenp; /* must be able to read the whole string in one go */ if (xlen < uio->uio_resid) return (EMSGSIZE); xlen = uio->uio_resid; if ((error = uiomove(buf, xlen, uio)) != 0) return (error); /* allow multiple writes without seeks */ uio->uio_offset = 0; /* cleanup string and remove trailing newline */ buf[xlen] = '\0'; xlen = strlen(buf); if (xlen > 0 && buf[xlen-1] == '\n') buf[--xlen] = '\0'; *buflenp = xlen; return (0); } vfs_namemap_t * vfs_findname(nm, buf, buflen) vfs_namemap_t *nm; char *buf; int buflen; { for (; nm->nm_name; nm++) if (bcmp(buf, nm->nm_name, buflen+1) == 0) return (nm); return (0); } #endif void linprocfs_exit(struct proc *p) { struct pfsnode *pfs; pid_t pid = p->p_pid; /* * The reason for this loop is not obvious -- basicly, * linprocfs_freevp(), which is called via vgone() (eventually), * removes the specified procfs node from the pfshead list. * It does this by *pfsp = pfs->pfs_next, meaning that it * overwrites the node. So when we do pfs = pfs->next, we * end up skipping the node that replaces the one that was * vgone'd. Since it may have been the last one on the list, * it may also have been set to null -- but *our* pfs pointer, * here, doesn't see this. So the loop starts from the beginning * again. * * This is not a for() loop because the final event * would be "pfs = pfs->pfs_next"; in the case where * pfs is set to pfshead again, that would mean that * pfshead is skipped over. * */ pfs = pfshead; while (pfs) { if (pfs->pfs_pid == pid) { vgone(PFSTOV(pfs)); pfs = pfshead; } else pfs = pfs->pfs_next; } } Index: head/sys/compat/linprocfs/linprocfs_vnops.c =================================================================== --- head/sys/compat/linprocfs/linprocfs_vnops.c (revision 60859) +++ head/sys/compat/linprocfs/linprocfs_vnops.c (revision 60860) @@ -1,994 +1,1023 @@ /* * Copyright (c) 1993, 1995 Jan-Simon Pendry * Copyright (c) 1993, 1995 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Jan-Simon Pendry. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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. * * @(#)procfs_vnops.c 8.18 (Berkeley) 5/21/95 * * $FreeBSD$ */ /* * procfs vnode interface */ #include #include #include #include #include #include #include #include #include #include #include #include +#include #include #include #include #include extern struct vnode *procfs_findtextvp __P((struct proc *)); extern int procfs_kmemaccess __P((struct proc *)); static int linprocfs_access __P((struct vop_access_args *)); static int linprocfs_badop __P((void)); static int linprocfs_bmap __P((struct vop_bmap_args *)); static int linprocfs_close __P((struct vop_close_args *)); static int linprocfs_getattr __P((struct vop_getattr_args *)); static int linprocfs_inactive __P((struct vop_inactive_args *)); static int linprocfs_ioctl __P((struct vop_ioctl_args *)); static int linprocfs_lookup __P((struct vop_lookup_args *)); static int linprocfs_open __P((struct vop_open_args *)); static int linprocfs_print __P((struct vop_print_args *)); static int linprocfs_readdir __P((struct vop_readdir_args *)); static int linprocfs_readlink __P((struct vop_readlink_args *)); static int linprocfs_reclaim __P((struct vop_reclaim_args *)); static int linprocfs_setattr __P((struct vop_setattr_args *)); /* * This is a list of the valid names in the * process-specific sub-directories. It is * used in linprocfs_lookup and linprocfs_readdir */ static struct proc_target { u_char pt_type; u_char pt_namlen; char *pt_name; pfstype pt_pfstype; int (*pt_valid) __P((struct proc *p)); } proc_targets[] = { #define N(s) sizeof(s)-1, s /* name type validp */ { DT_DIR, N("."), Pproc, NULL }, { DT_DIR, N(".."), Proot, NULL }, - { DT_REG, N("exe"), Pexe, linprocfs_validfile }, { DT_REG, N("mem"), Pmem, NULL }, + { DT_LNK, N("exe"), Pexe, NULL }, #undef N }; static const int nproc_targets = sizeof(proc_targets) / sizeof(proc_targets[0]); static pid_t atopid __P((const char *, u_int)); /* * set things up for doing i/o on * the pfsnode (vp). (vp) is locked * on entry, and should be left locked * on exit. * * for procfs we don't need to do anything * in particular for i/o. all that is done * is to support exclusive open on process * memory images. */ static int linprocfs_open(ap) struct vop_open_args /* { struct vnode *a_vp; int a_mode; struct ucred *a_cred; struct proc *a_p; } */ *ap; { struct pfsnode *pfs = VTOPFS(ap->a_vp); struct proc *p1, *p2; p2 = PFIND(pfs->pfs_pid); if (p2 == NULL) return (ENOENT); if (pfs->pfs_pid && !PRISON_CHECK(ap->a_p, p2)) return (ENOENT); switch (pfs->pfs_type) { case Pmem: if (((pfs->pfs_flags & FWRITE) && (ap->a_mode & O_EXCL)) || ((pfs->pfs_flags & O_EXCL) && (ap->a_mode & FWRITE))) return (EBUSY); p1 = ap->a_p; if (p_trespass(p1, p2) && !procfs_kmemaccess(p1)) return (EPERM); if (ap->a_mode & FWRITE) pfs->pfs_flags = ap->a_mode & (FWRITE|O_EXCL); return (0); default: break; } return (0); } /* * close the pfsnode (vp) after doing i/o. * (vp) is not locked on entry or exit. * * nothing to do for procfs other than undo * any exclusive open flag (see _open above). */ static int linprocfs_close(ap) struct vop_close_args /* { struct vnode *a_vp; int a_fflag; struct ucred *a_cred; struct proc *a_p; } */ *ap; { struct pfsnode *pfs = VTOPFS(ap->a_vp); struct proc *p; switch (pfs->pfs_type) { case Pmem: if ((ap->a_fflag & FWRITE) && (pfs->pfs_flags & O_EXCL)) pfs->pfs_flags &= ~(FWRITE|O_EXCL); /* * This rather complicated-looking code is trying to * determine if this was the last close on this particular * vnode. While one would expect v_usecount to be 1 at * that point, it seems that (according to John Dyson) * the VM system will bump up the usecount. So: if the * usecount is 2, and VOBJBUF is set, then this is really * the last close. Otherwise, if the usecount is < 2 * then it is definitely the last close. * If this is the last close, then it checks to see if * the target process has PF_LINGER set in p_pfsflags, * if this is *not* the case, then the process' stop flags * are cleared, and the process is woken up. This is * to help prevent the case where a process has been * told to stop on an event, but then the requesting process * has gone away or forgotten about it. */ if ((ap->a_vp->v_usecount < 2) && (p = pfind(pfs->pfs_pid)) && !(p->p_pfsflags & PF_LINGER)) { p->p_stops = 0; p->p_step = 0; wakeup(&p->p_step); } break; default: break; } return (0); } /* * do an ioctl operation on a pfsnode (vp). * (vp) is not locked on entry or exit. */ static int linprocfs_ioctl(ap) struct vop_ioctl_args *ap; { struct pfsnode *pfs = VTOPFS(ap->a_vp); struct proc *procp, *p; int error; int signo; struct procfs_status *psp; unsigned char flags; p = ap->a_p; procp = pfind(pfs->pfs_pid); if (procp == NULL) { return ENOTTY; } if (p_trespass(p, procp)) return EPERM; switch (ap->a_command) { case PIOCBIS: procp->p_stops |= *(unsigned int*)ap->a_data; break; case PIOCBIC: procp->p_stops &= ~*(unsigned int*)ap->a_data; break; case PIOCSFL: /* * NFLAGS is "non-suser_xxx flags" -- currently, only * PFS_ISUGID ("ignore set u/g id"); */ #define NFLAGS (PF_ISUGID) flags = (unsigned char)*(unsigned int*)ap->a_data; if (flags & NFLAGS && (error = suser(p))) return error; procp->p_pfsflags = flags; break; case PIOCGFL: *(unsigned int*)ap->a_data = (unsigned int)procp->p_pfsflags; case PIOCSTATUS: psp = (struct procfs_status *)ap->a_data; psp->state = (procp->p_step == 0); psp->flags = procp->p_pfsflags; psp->events = procp->p_stops; if (procp->p_step) { psp->why = procp->p_stype; psp->val = procp->p_xstat; } else { psp->why = psp->val = 0; /* Not defined values */ } break; case PIOCWAIT: psp = (struct procfs_status *)ap->a_data; if (procp->p_step == 0) { error = tsleep(&procp->p_stype, PWAIT | PCATCH, "piocwait", 0); if (error) return error; } psp->state = 1; /* It stopped */ psp->flags = procp->p_pfsflags; psp->events = procp->p_stops; psp->why = procp->p_stype; /* why it stopped */ psp->val = procp->p_xstat; /* any extra info */ break; case PIOCCONT: /* Restart a proc */ if (procp->p_step == 0) return EINVAL; /* Can only start a stopped process */ if ((signo = *(int*)ap->a_data) != 0) { if (signo >= NSIG || signo <= 0) return EINVAL; psignal(procp, signo); } procp->p_step = 0; wakeup(&procp->p_step); break; default: return (ENOTTY); } return 0; } /* * do block mapping for pfsnode (vp). * since we don't use the buffer cache * for procfs this function should never * be called. in any case, it's not clear * what part of the kernel ever makes use * of this function. for sanity, this is the * usual no-op bmap, although returning * (EIO) would be a reasonable alternative. */ static int linprocfs_bmap(ap) struct vop_bmap_args /* { struct vnode *a_vp; daddr_t a_bn; struct vnode **a_vpp; daddr_t *a_bnp; int *a_runp; } */ *ap; { if (ap->a_vpp != NULL) *ap->a_vpp = ap->a_vp; if (ap->a_bnp != NULL) *ap->a_bnp = ap->a_bn; if (ap->a_runp != NULL) *ap->a_runp = 0; return (0); } /* * linprocfs_inactive is called when the pfsnode * is vrele'd and the reference count goes * to zero. (vp) will be on the vnode free * list, so to get it back vget() must be * used. * * (vp) is locked on entry, but must be unlocked on exit. */ static int linprocfs_inactive(ap) struct vop_inactive_args /* { struct vnode *a_vp; } */ *ap; { struct vnode *vp = ap->a_vp; VOP_UNLOCK(vp, 0, ap->a_p); return (0); } /* * _reclaim is called when getnewvnode() * wants to make use of an entry on the vnode * free list. at this time the filesystem needs * to free any private data and remove the node * from any private lists. */ static int linprocfs_reclaim(ap) struct vop_reclaim_args /* { struct vnode *a_vp; } */ *ap; { return (linprocfs_freevp(ap->a_vp)); } /* * _print is used for debugging. * just print a readable description * of (vp). */ static int linprocfs_print(ap) struct vop_print_args /* { struct vnode *a_vp; } */ *ap; { struct pfsnode *pfs = VTOPFS(ap->a_vp); printf("tag VT_PROCFS, type %d, pid %ld, mode %x, flags %lx\n", pfs->pfs_type, (long)pfs->pfs_pid, pfs->pfs_mode, pfs->pfs_flags); return (0); } /* * generic entry point for unsupported operations */ static int linprocfs_badop() { return (EIO); } /* * Invent attributes for pfsnode (vp) and store * them in (vap). * Directories lengths are returned as zero since * any real length would require the genuine size * to be computed, and nothing cares anyway. * * this is relatively minimal for procfs. */ static int linprocfs_getattr(ap) struct vop_getattr_args /* { struct vnode *a_vp; struct vattr *a_vap; struct ucred *a_cred; struct proc *a_p; } */ *ap; { struct pfsnode *pfs = VTOPFS(ap->a_vp); struct vattr *vap = ap->a_vap; struct proc *procp; int error; /* * First make sure that the process and its credentials * still exist. */ switch (pfs->pfs_type) { case Proot: case Pself: procp = 0; break; default: procp = PFIND(pfs->pfs_pid); if (procp == 0 || procp->p_cred == NULL || procp->p_ucred == NULL) return (ENOENT); } error = 0; /* start by zeroing out the attributes */ VATTR_NULL(vap); /* next do all the common fields */ vap->va_type = ap->a_vp->v_type; vap->va_mode = pfs->pfs_mode; vap->va_fileid = pfs->pfs_fileno; vap->va_flags = 0; vap->va_blocksize = PAGE_SIZE; vap->va_bytes = vap->va_size = 0; vap->va_fsid = ap->a_vp->v_mount->mnt_stat.f_fsid.val[0]; /* * Make all times be current TOD. * It would be possible to get the process start * time from the p_stat structure, but there's * no "file creation" time stamp anyway, and the * p_stat structure is not addressible if u. gets * swapped out for that process. */ nanotime(&vap->va_ctime); vap->va_atime = vap->va_mtime = vap->va_ctime; /* * If the process has exercised some setuid or setgid * privilege, then rip away read/write permission so * that only root can gain access. */ switch (pfs->pfs_type) { case Pmem: /* Retain group kmem readablity. */ if (procp->p_flag & P_SUGID) vap->va_mode &= ~(VREAD|VWRITE); break; default: break; } /* * now do the object specific fields * * The size could be set from struct reg, but it's hardly * worth the trouble, and it puts some (potentially) machine * dependent data into this machine-independent code. If it * becomes important then this function should break out into * a per-file stat function in the corresponding .c file. */ vap->va_nlink = 1; if (procp) { vap->va_uid = procp->p_ucred->cr_uid; vap->va_gid = procp->p_ucred->cr_gid; } switch (pfs->pfs_type) { case Proot: /* * Set nlink to 1 to tell fts(3) we don't actually know. */ vap->va_nlink = 1; vap->va_uid = 0; vap->va_gid = 0; vap->va_size = vap->va_bytes = DEV_BSIZE; break; case Pself: { char buf[16]; /* should be enough */ vap->va_uid = 0; vap->va_gid = 0; vap->va_size = vap->va_bytes = snprintf(buf, sizeof(buf), "%ld", (long)curproc->p_pid); break; } case Pproc: vap->va_nlink = nproc_targets; vap->va_size = vap->va_bytes = DEV_BSIZE; break; - case Pexe: - error = EOPNOTSUPP; + case Pexe: { + char *fullpath, *freepath; + error = textvp_fullpath(procp, &fullpath, &freepath); + if (error == 0) { + vap->va_size = strlen(fullpath); + free(freepath, M_TEMP); + } else { + vap->va_size = sizeof("unknown") - 1; + error = 0; + } + vap->va_bytes = vap->va_size; break; + } case Pmeminfo: case Pcpuinfo: break; case Pmem: /* * If we denied owner access earlier, then we have to * change the owner to root - otherwise 'ps' and friends * will break even though they are setgid kmem. *SIGH* */ if (procp->p_flag & P_SUGID) vap->va_uid = 0; else vap->va_uid = procp->p_ucred->cr_uid; vap->va_gid = KMEM_GROUP; break; default: panic("linprocfs_getattr"); } return (error); } static int linprocfs_setattr(ap) struct vop_setattr_args /* { struct vnode *a_vp; struct vattr *a_vap; struct ucred *a_cred; struct proc *a_p; } */ *ap; { if (ap->a_vap->va_flags != VNOVAL) return (EOPNOTSUPP); /* * just fake out attribute setting * it's not good to generate an error * return, otherwise things like creat() * will fail when they try to set the * file length to 0. worse, this means * that echo $note > /proc/$pid/note will fail. */ return (0); } /* * implement access checking. * * something very similar to this code is duplicated * throughout the 4bsd kernel and should be moved * into kern/vfs_subr.c sometime. * * actually, the check for super-user is slightly * broken since it will allow read access to write-only * objects. this doesn't cause any particular trouble * but does mean that the i/o entry points need to check * that the operation really does make sense. */ static int linprocfs_access(ap) struct vop_access_args /* { struct vnode *a_vp; int a_mode; struct ucred *a_cred; struct proc *a_p; } */ *ap; { struct vattr *vap; struct vattr vattr; int error; /* * If you're the super-user, * you always get access. */ if (ap->a_cred->cr_uid == 0) return (0); vap = &vattr; error = VOP_GETATTR(ap->a_vp, vap, ap->a_cred, ap->a_p); if (error) return (error); /* * Access check is based on only one of owner, group, public. * If not owner, then check group. If not a member of the * group, then check public access. */ if (ap->a_cred->cr_uid != vap->va_uid) { gid_t *gp; int i; ap->a_mode >>= 3; gp = ap->a_cred->cr_groups; for (i = 0; i < ap->a_cred->cr_ngroups; i++, gp++) if (vap->va_gid == *gp) goto found; ap->a_mode >>= 3; found: ; } if ((vap->va_mode & ap->a_mode) == ap->a_mode) return (0); return (EACCES); } /* * lookup. this is incredibly complicated in the * general case, however for most pseudo-filesystems * very little needs to be done. * * unless you want to get a migraine, just make sure your * filesystem doesn't do any locking of its own. otherwise * read and inwardly digest ufs_lookup(). */ static int linprocfs_lookup(ap) struct vop_lookup_args /* { struct vnode * a_dvp; struct vnode ** a_vpp; struct componentname * a_cnp; } */ *ap; { struct componentname *cnp = ap->a_cnp; struct vnode **vpp = ap->a_vpp; struct vnode *dvp = ap->a_dvp; char *pname = cnp->cn_nameptr; - struct proc *curp = cnp->cn_proc; struct proc_target *pt; - struct vnode *fvp; pid_t pid; struct pfsnode *pfs; struct proc *p; int i; *vpp = NULL; if (cnp->cn_nameiop == DELETE || cnp->cn_nameiop == RENAME) return (EROFS); if (cnp->cn_namelen == 1 && *pname == '.') { *vpp = dvp; VREF(dvp); /* vn_lock(dvp, LK_EXCLUSIVE | LK_RETRY, curp); */ return (0); } pfs = VTOPFS(dvp); switch (pfs->pfs_type) { case Proot: if (cnp->cn_flags & ISDOTDOT) return (EIO); if (CNEQ(cnp, "self", 4)) return (linprocfs_allocvp(dvp->v_mount, vpp, 0, Pself)); if (CNEQ(cnp, "meminfo", 7)) return (linprocfs_allocvp(dvp->v_mount, vpp, 0, Pmeminfo)); if (CNEQ(cnp, "cpuinfo", 7)) return (linprocfs_allocvp(dvp->v_mount, vpp, 0, Pcpuinfo)); pid = atopid(pname, cnp->cn_namelen); if (pid == NO_PID) break; p = PFIND(pid); if (p == 0) break; return (linprocfs_allocvp(dvp->v_mount, vpp, pid, Pproc)); case Pproc: if (cnp->cn_flags & ISDOTDOT) return (linprocfs_root(dvp->v_mount, vpp)); p = PFIND(pfs->pfs_pid); if (p == 0) break; for (pt = proc_targets, i = 0; i < nproc_targets; pt++, i++) { if (cnp->cn_namelen == pt->pt_namlen && bcmp(pt->pt_name, pname, cnp->cn_namelen) == 0 && (pt->pt_valid == NULL || (*pt->pt_valid)(p))) goto found; } break; found: - if (pt->pt_pfstype == Pexe) { - fvp = procfs_findtextvp(p); - /* We already checked that it exists. */ - VREF(fvp); - vn_lock(fvp, LK_EXCLUSIVE | LK_RETRY, curp); - *vpp = fvp; - return (0); - } - return (linprocfs_allocvp(dvp->v_mount, vpp, pfs->pfs_pid, pt->pt_pfstype)); default: return (ENOTDIR); } return (cnp->cn_nameiop == LOOKUP ? ENOENT : EROFS); } /* * Does this process have a text file? */ int linprocfs_validfile(p) struct proc *p; { return (procfs_findtextvp(p) != NULLVP); } /* * readdir() returns directory entries from pfsnode (vp). * * We generate just one directory entry at a time, as it would probably * not pay off to buffer several entries locally to save uiomove calls. */ static int linprocfs_readdir(ap) struct vop_readdir_args /* { struct vnode *a_vp; struct uio *a_uio; struct ucred *a_cred; int *a_eofflag; int *a_ncookies; u_long **a_cookies; } */ *ap; { struct uio *uio = ap->a_uio; struct dirent d; struct dirent *dp = &d; struct pfsnode *pfs; int count, error, i, off; static u_int delen; if (!delen) { d.d_namlen = PROCFS_NAMELEN; delen = GENERIC_DIRSIZ(&d); } pfs = VTOPFS(ap->a_vp); off = (int)uio->uio_offset; if (off != uio->uio_offset || off < 0 || off % delen != 0 || uio->uio_resid < delen) return (EINVAL); error = 0; count = 0; i = off / delen; switch (pfs->pfs_type) { /* * this is for the process-specific sub-directories. * all that is needed to is copy out all the entries * from the procent[] table (top of this file). */ case Pproc: { struct proc *p; struct proc_target *pt; p = PFIND(pfs->pfs_pid); if (p == NULL) break; if (!PRISON_CHECK(curproc, p)) break; for (pt = &proc_targets[i]; uio->uio_resid >= delen && i < nproc_targets; pt++, i++) { if (pt->pt_valid && (*pt->pt_valid)(p) == 0) continue; dp->d_reclen = delen; dp->d_fileno = PROCFS_FILENO(pfs->pfs_pid, pt->pt_pfstype); dp->d_namlen = pt->pt_namlen; bcopy(pt->pt_name, dp->d_name, pt->pt_namlen + 1); dp->d_type = pt->pt_type; if ((error = uiomove((caddr_t)dp, delen, uio)) != 0) break; } break; } /* * this is for the root of the procfs filesystem * what is needed is a special entry for "self" * followed by an entry for each process on allproc #ifdef PROCFS_ZOMBIE * and zombproc. #endif */ case Proot: { #ifdef PROCFS_ZOMBIE int doingzomb = 0; #endif int pcnt = 0; volatile struct proc *p = allproc.lh_first; for (; p && uio->uio_resid >= delen; i++, pcnt++) { bzero((char *) dp, delen); dp->d_reclen = delen; switch (i) { case 0: /* `.' */ case 1: /* `..' */ dp->d_fileno = PROCFS_FILENO(0, Proot); dp->d_namlen = i + 1; bcopy("..", dp->d_name, dp->d_namlen); dp->d_name[i + 1] = '\0'; dp->d_type = DT_DIR; break; case 2: dp->d_fileno = PROCFS_FILENO(0, Pself); dp->d_namlen = 4; bcopy("self", dp->d_name, 5); dp->d_type = DT_LNK; break; case 3: dp->d_fileno = PROCFS_FILENO(0, Pmeminfo); dp->d_namlen = 7; bcopy("meminfo", dp->d_name, 8); dp->d_type = DT_REG; break; case 4: dp->d_fileno = PROCFS_FILENO(0, Pcpuinfo); dp->d_namlen = 7; bcopy("cpuinfo", dp->d_name, 8); dp->d_type = DT_REG; break; default: while (pcnt < i) { p = p->p_list.le_next; if (!p) goto done; if (!PRISON_CHECK(curproc, p)) continue; pcnt++; } while (!PRISON_CHECK(curproc, p)) { p = p->p_list.le_next; if (!p) goto done; } dp->d_fileno = PROCFS_FILENO(p->p_pid, Pproc); dp->d_namlen = sprintf(dp->d_name, "%ld", (long)p->p_pid); dp->d_type = DT_REG; p = p->p_list.le_next; break; } if ((error = uiomove((caddr_t)dp, delen, uio)) != 0) break; } done: #ifdef PROCFS_ZOMBIE if (p == 0 && doingzomb == 0) { doingzomb = 1; p = zombproc.lh_first; goto again; } #endif break; } default: error = ENOTDIR; break; } uio->uio_offset = i * delen; return (error); } /* - * readlink reads the link of `self' + * readlink reads the link of `self' or `exe' */ static int linprocfs_readlink(ap) struct vop_readlink_args *ap; { char buf[16]; /* should be enough */ - int len; + struct proc *procp; + struct vnode *vp = ap->a_vp; + struct pfsnode *pfs = VTOPFS(vp); + char *fullpath, *freepath; + int error, len; - if (VTOPFS(ap->a_vp)->pfs_fileno != PROCFS_FILENO(0, Pself)) - return (EINVAL); + switch (pfs->pfs_type) { + case Pself: + if (pfs->pfs_fileno != PROCFS_FILENO(0, Pself)) + return (EINVAL); - len = snprintf(buf, sizeof(buf), "%ld", (long)curproc->p_pid); + len = snprintf(buf, sizeof(buf), "%ld", (long)curproc->p_pid); - return (uiomove((caddr_t)buf, len, ap->a_uio)); + return (uiomove(buf, len, ap->a_uio)); + /* + * There _should_ be no way for an entire process to disappear + * from under us... + */ + case Pexe: + procp = PFIND(pfs->pfs_pid); + if (procp == NULL || procp->p_cred == NULL || + procp->p_ucred == NULL) { + printf("linprocfs_readlink: pid %d disappeared\n", + pfs->pfs_pid); + return (uiomove("unknown", sizeof("unknown") - 1, + ap->a_uio)); + } + error = textvp_fullpath(procp, &fullpath, &freepath); + if (error != 0) + return (uiomove("unknown", sizeof("unknown") - 1, + ap->a_uio)); + error = uiomove(fullpath, strlen(fullpath), ap->a_uio); + free(freepath, M_TEMP); + return (error); + default: + return (EINVAL); + } } /* * convert decimal ascii to pid_t */ static pid_t atopid(b, len) const char *b; u_int len; { pid_t p = 0; while (len--) { char c = *b++; if (c < '0' || c > '9') return (NO_PID); p = 10 * p + (c - '0'); if (p > PID_MAX) return (NO_PID); } return (p); } /* * procfs vnode operations. */ vop_t **linprocfs_vnodeop_p; static struct vnodeopv_entry_desc linprocfs_vnodeop_entries[] = { { &vop_default_desc, (vop_t *) vop_defaultop }, { &vop_access_desc, (vop_t *) linprocfs_access }, { &vop_advlock_desc, (vop_t *) linprocfs_badop }, { &vop_bmap_desc, (vop_t *) linprocfs_bmap }, { &vop_close_desc, (vop_t *) linprocfs_close }, { &vop_create_desc, (vop_t *) linprocfs_badop }, { &vop_getattr_desc, (vop_t *) linprocfs_getattr }, { &vop_inactive_desc, (vop_t *) linprocfs_inactive }, { &vop_link_desc, (vop_t *) linprocfs_badop }, { &vop_lookup_desc, (vop_t *) linprocfs_lookup }, { &vop_mkdir_desc, (vop_t *) linprocfs_badop }, { &vop_mknod_desc, (vop_t *) linprocfs_badop }, { &vop_open_desc, (vop_t *) linprocfs_open }, { &vop_pathconf_desc, (vop_t *) vop_stdpathconf }, { &vop_print_desc, (vop_t *) linprocfs_print }, { &vop_read_desc, (vop_t *) linprocfs_rw }, { &vop_readdir_desc, (vop_t *) linprocfs_readdir }, { &vop_readlink_desc, (vop_t *) linprocfs_readlink }, { &vop_reclaim_desc, (vop_t *) linprocfs_reclaim }, { &vop_remove_desc, (vop_t *) linprocfs_badop }, { &vop_rename_desc, (vop_t *) linprocfs_badop }, { &vop_rmdir_desc, (vop_t *) linprocfs_badop }, { &vop_setattr_desc, (vop_t *) linprocfs_setattr }, { &vop_symlink_desc, (vop_t *) linprocfs_badop }, { &vop_write_desc, (vop_t *) linprocfs_rw }, { &vop_ioctl_desc, (vop_t *) linprocfs_ioctl }, { NULL, NULL } }; static struct vnodeopv_desc linprocfs_vnodeop_opv_desc = { &linprocfs_vnodeop_p, linprocfs_vnodeop_entries }; VNODEOP_SET(linprocfs_vnodeop_opv_desc); Index: head/sys/i386/linux/linprocfs/linprocfs.h =================================================================== --- head/sys/i386/linux/linprocfs/linprocfs.h (revision 60859) +++ head/sys/i386/linux/linprocfs/linprocfs.h (revision 60860) @@ -1,135 +1,134 @@ /* * Copyright (c) 1993 Jan-Simon Pendry * Copyright (c) 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Jan-Simon Pendry. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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. * * @(#)procfs.h 8.9 (Berkeley) 5/14/95 * * $FreeBSD$ */ /* * The different types of node in a procfs filesystem */ typedef enum { Proot, /* the filesystem root */ Pself, /* symbolic link for curproc */ Pproc, /* a process-specific sub-directory */ Pexe, /* the executable file */ Pmem, /* the process's memory image */ Pmeminfo, Pcpuinfo } pfstype; /* * control data for the proc file system. */ struct pfsnode { struct pfsnode *pfs_next; /* next on list */ struct vnode *pfs_vnode; /* vnode associated with this pfsnode */ pfstype pfs_type; /* type of procfs node */ pid_t pfs_pid; /* associated process */ u_short pfs_mode; /* mode bits for stat() */ u_long pfs_flags; /* open flags */ u_long pfs_fileno; /* unique file id */ pid_t pfs_lockowner; /* pfs lock owner */ }; #define PROCFS_NAMELEN 8 /* max length of a filename component */ /* * Kernel stuff follows */ #ifdef _KERNEL #define CNEQ(cnp, s, len) \ ((cnp)->cn_namelen == (len) && \ (bcmp((s), (cnp)->cn_nameptr, (len)) == 0)) #define KMEM_GROUP 2 #define PROCFS_FILENO(pid, type) \ (((type) < Pproc) ? \ ((type) + 2) : \ ((((pid)+1) << 4) + ((int) (type)))) /* * Convert between pfsnode vnode */ #define VTOPFS(vp) ((struct pfsnode *)(vp)->v_data) #define PFSTOV(pfs) ((pfs)->pfs_vnode) typedef struct vfs_namemap vfs_namemap_t; struct vfs_namemap { const char *nm_name; int nm_val; }; int vfs_getuserstr __P((struct uio *, char *, int *)); vfs_namemap_t *vfs_findname __P((vfs_namemap_t *, char *, int)); /* */ struct reg; struct fpreg; struct dbreg; #define PFIND(pid) ((pid) ? pfind(pid) : &proc0) void linprocfs_exit __P((struct proc *)); int linprocfs_freevp __P((struct vnode *)); int linprocfs_allocvp __P((struct mount *, struct vnode **, long, pfstype)); -struct vnode *linprocfs_findtextvp __P((struct proc *)); int linprocfs_sstep __P((struct proc *)); void linprocfs_fix_sstep __P((struct proc *)); #if 0 int linprocfs_read_regs __P((struct proc *, struct reg *)); int linprocfs_write_regs __P((struct proc *, struct reg *)); int linprocfs_read_fpregs __P((struct proc *, struct fpreg *)); int linprocfs_write_fpregs __P((struct proc *, struct fpreg *)); int linprocfs_read_dbregs __P((struct proc *, struct dbreg *)); int linprocfs_write_dbregs __P((struct proc *, struct dbreg *)); #endif int linprocfs_domeminfo __P((struct proc *, struct proc *, struct pfsnode *pfsp, struct uio *uio)); int linprocfs_docpuinfo __P((struct proc *, struct proc *, struct pfsnode *pfsp, struct uio *uio)); /* functions to check whether or not files should be displayed */ int linprocfs_validfile __P((struct proc *)); #define PROCFS_LOCKED 0x01 #define PROCFS_WANT 0x02 extern vop_t **linprocfs_vnodeop_p; int linprocfs_root __P((struct mount *, struct vnode **)); int linprocfs_rw __P((struct vop_read_args *)); #endif /* _KERNEL */ Index: head/sys/i386/linux/linprocfs/linprocfs_misc.c =================================================================== --- head/sys/i386/linux/linprocfs/linprocfs_misc.c (revision 60859) +++ head/sys/i386/linux/linprocfs/linprocfs_misc.c (revision 60860) @@ -1,202 +1,209 @@ /* * Copyright (c) 1993 Jan-Simon Pendry * Copyright (c) 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Jan-Simon Pendry. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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. * * @(#)procfs_status.c 8.4 (Berkeley) 6/15/94 * * $FreeBSD$ */ #include #include #include #include #include #include #include #include #include #include #include #include +#include #include #include #include #include #include struct proc; int linprocfs_domeminfo(curp, p, pfs, uio) struct proc *curp; struct proc *p; struct pfsnode *pfs; struct uio *uio; { char *ps; int xlen; int error; char psbuf[512]; /* XXX - conservative */ unsigned long memtotal; /* total memory in bytes */ unsigned long memused; /* used memory in bytes */ unsigned long memfree; /* free memory in bytes */ unsigned long memshared; /* shared memory ??? */ unsigned long buffers, cached; /* buffer / cache memory ??? */ unsigned long swaptotal; /* total swap space in bytes */ unsigned long swapused; /* used swap space in bytes */ unsigned long swapfree; /* free swap space in bytes */ + vm_object_t object; if (uio->uio_rw != UIO_READ) return (EOPNOTSUPP); memtotal = physmem * PAGE_SIZE; /* * The correct thing here would be: * memfree = cnt.v_free_count * PAGE_SIZE; memused = memtotal - memfree; * * but it might mislead linux binaries into thinking there * is very little memory left, so we cheat and tell them that * all memory that isn't wired down is free. */ memused = cnt.v_wire_count * PAGE_SIZE; memfree = memtotal - memused; swaptotal = swapblist->bl_blocks * 1024; /* XXX why 1024? */ swapfree = swapblist->bl_root->u.bmu_avail * PAGE_SIZE; swapused = swaptotal - swapfree; - memshared = 0; /* XXX what's this supposed to be? */ + memshared = 0; + for (object = TAILQ_FIRST(&vm_object_list); object != NULL; + object = TAILQ_NEXT(object, object_list)) + if (object->shadow_count > 1) + memshared += object->resident_page_count; + memshared *= PAGE_SIZE; /* * We'd love to be able to write: * buffers = bufspace; * * but bufspace is internal to vfs_bio.c and we don't feel * like unstaticizing it just for linprocfs's sake. */ buffers = 0; cached = cnt.v_cache_count * PAGE_SIZE; ps = psbuf; ps += sprintf(ps, " total: used: free: shared: buffers: cached:\n" "Mem: %lu %lu %lu %lu %lu %lu\n" "Swap: %lu %lu %lu\n" "MemTotal: %9lu kB\n" "MemFree: %9lu kB\n" "MemShared:%9lu kB\n" "Buffers: %9lu kB\n" "Cached: %9lu kB\n" "SwapTotal:%9lu kB\n" "SwapFree: %9lu kB\n", memtotal, memused, memfree, memshared, buffers, cached, swaptotal, swapused, swapfree, memtotal >> 10, memfree >> 10, memshared >> 10, buffers >> 10, cached >> 10, swaptotal >> 10, swapfree >> 10); xlen = ps - psbuf; xlen -= uio->uio_offset; ps = psbuf + uio->uio_offset; xlen = imin(xlen, uio->uio_resid); if (xlen <= 0) error = 0; else error = uiomove(ps, xlen, uio); return (error); } int linprocfs_docpuinfo(curp, p, pfs, uio) struct proc *curp; struct proc *p; struct pfsnode *pfs; struct uio *uio; { char *ps; int xlen; int error; char psbuf[512]; /* XXX - conservative */ char *class; #if 0 extern char *cpu_model; /* Yuck */ #endif if (uio->uio_rw != UIO_READ) return (EOPNOTSUPP); switch (cpu_class) { case CPUCLASS_286: class = "286"; break; case CPUCLASS_386: class = "386"; break; case CPUCLASS_486: class = "486"; break; case CPUCLASS_586: class = "586"; break; case CPUCLASS_686: class = "686"; break; default: class = "unknown"; break; } ps = psbuf; ps += sprintf(ps, "processor : %d\n" "cpu : %.3s\n" "model : %.20s\n" "vendor_id : %.20s\n" "stepping : %d\n", 0, class, "unknown", cpu_vendor, cpu_id); xlen = ps - psbuf; xlen -= uio->uio_offset; ps = psbuf + uio->uio_offset; xlen = imin(xlen, uio->uio_resid); if (xlen <= 0) error = 0; else error = uiomove(ps, xlen, uio); return (error); } Index: head/sys/i386/linux/linprocfs/linprocfs_subr.c =================================================================== --- head/sys/i386/linux/linprocfs/linprocfs_subr.c (revision 60859) +++ head/sys/i386/linux/linprocfs/linprocfs_subr.c (revision 60860) @@ -1,353 +1,359 @@ /* * Copyright (c) 1993 Jan-Simon Pendry * Copyright (c) 1993 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Jan-Simon Pendry. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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. * * @(#)procfs_subr.c 8.6 (Berkeley) 5/14/95 * * $FreeBSD$ */ #include #include #include #include #include #include static struct pfsnode *pfshead; static int pfsvplock; extern int procfs_domem __P((struct proc *, struct proc *, struct pfsnode *pfsp, struct uio *uio)); /* * allocate a pfsnode/vnode pair. the vnode is * referenced, but not locked. * * the pid, pfs_type, and mount point uniquely * identify a pfsnode. the mount point is needed * because someone might mount this filesystem * twice. * * all pfsnodes are maintained on a singly-linked * list. new nodes are only allocated when they cannot * be found on this list. entries on the list are * removed when the vfs reclaim entry is called. * * a single lock is kept for the entire list. this is * needed because the getnewvnode() function can block * waiting for a vnode to become free, in which case there * may be more than one process trying to get the same * vnode. this lock is only taken if we are going to * call getnewvnode, since the kernel itself is single-threaded. * * if an entry is found on the list, then call vget() to * take a reference. this is done because there may be * zero references to it and so it needs to removed from * the vnode free list. */ int linprocfs_allocvp(mp, vpp, pid, pfs_type) struct mount *mp; struct vnode **vpp; long pid; pfstype pfs_type; { struct proc *p = curproc; /* XXX */ struct pfsnode *pfs; struct vnode *vp; struct pfsnode **pp; int error; loop: for (pfs = pfshead; pfs != 0; pfs = pfs->pfs_next) { vp = PFSTOV(pfs); if (pfs->pfs_pid == pid && pfs->pfs_type == pfs_type && vp->v_mount == mp) { if (vget(vp, 0, p)) goto loop; *vpp = vp; return (0); } } /* * otherwise lock the vp list while we call getnewvnode * since that can block. */ if (pfsvplock & PROCFS_LOCKED) { pfsvplock |= PROCFS_WANT; (void) tsleep((caddr_t) &pfsvplock, PINOD, "pfsavp", 0); goto loop; } pfsvplock |= PROCFS_LOCKED; /* * Do the MALLOC before the getnewvnode since doing so afterward * might cause a bogus v_data pointer to get dereferenced * elsewhere if MALLOC should block. */ MALLOC(pfs, struct pfsnode *, sizeof(struct pfsnode), M_TEMP, M_WAITOK); if ((error = getnewvnode(VT_PROCFS, mp, linprocfs_vnodeop_p, vpp)) != 0) { FREE(pfs, M_TEMP); goto out; } vp = *vpp; vp->v_data = pfs; pfs->pfs_next = 0; pfs->pfs_pid = (pid_t) pid; pfs->pfs_type = pfs_type; pfs->pfs_vnode = vp; pfs->pfs_flags = 0; pfs->pfs_lockowner = 0; pfs->pfs_fileno = PROCFS_FILENO(pid, pfs_type); switch (pfs_type) { case Proot: /* /proc = dr-xr-xr-x */ pfs->pfs_mode = (VREAD|VEXEC) | (VREAD|VEXEC) >> 3 | (VREAD|VEXEC) >> 6; vp->v_type = VDIR; vp->v_flag = VROOT; break; case Pself: /* /proc/self = lr--r--r-- */ pfs->pfs_mode = (VREAD) | (VREAD >> 3) | (VREAD >> 6); vp->v_type = VLNK; break; case Pproc: pfs->pfs_mode = (VREAD|VEXEC) | (VREAD|VEXEC) >> 3 | (VREAD|VEXEC) >> 6; vp->v_type = VDIR; break; case Pexe: + pfs->pfs_mode = (VREAD|VEXEC) | + (VREAD|VEXEC) >> 3 | + (VREAD|VEXEC) >> 6; + vp->v_type = VLNK; + break; + case Pmem: pfs->pfs_mode = (VREAD|VWRITE) | (VREAD) >> 3;; vp->v_type = VREG; break; case Pmeminfo: case Pcpuinfo: pfs->pfs_mode = (VREAD) | (VREAD >> 3) | (VREAD >> 6); vp->v_type = VREG; break; default: panic("linprocfs_allocvp"); } /* add to procfs vnode list */ for (pp = &pfshead; *pp; pp = &(*pp)->pfs_next) continue; *pp = pfs; out: pfsvplock &= ~PROCFS_LOCKED; if (pfsvplock & PROCFS_WANT) { pfsvplock &= ~PROCFS_WANT; wakeup((caddr_t) &pfsvplock); } return (error); } int linprocfs_freevp(vp) struct vnode *vp; { struct pfsnode **pfspp; struct pfsnode *pfs = VTOPFS(vp); for (pfspp = &pfshead; *pfspp != 0; pfspp = &(*pfspp)->pfs_next) { if (*pfspp == pfs) { *pfspp = pfs->pfs_next; break; } } FREE(vp->v_data, M_TEMP); vp->v_data = 0; return (0); } int linprocfs_rw(ap) struct vop_read_args *ap; { struct vnode *vp = ap->a_vp; struct uio *uio = ap->a_uio; struct proc *curp = uio->uio_procp; struct pfsnode *pfs = VTOPFS(vp); struct proc *p; int rtval; p = PFIND(pfs->pfs_pid); if (p == 0) return (EINVAL); if (p->p_pid == 1 && securelevel > 0 && uio->uio_rw == UIO_WRITE) return (EACCES); while (pfs->pfs_lockowner) { tsleep(&pfs->pfs_lockowner, PRIBIO, "pfslck", 0); } pfs->pfs_lockowner = curproc->p_pid; switch (pfs->pfs_type) { case Pmem: rtval = procfs_domem(curp, p, pfs, uio); break; case Pmeminfo: rtval = linprocfs_domeminfo(curp, p, pfs, uio); break; case Pcpuinfo: rtval = linprocfs_docpuinfo(curp, p, pfs, uio); break; default: rtval = EOPNOTSUPP; break; } pfs->pfs_lockowner = 0; wakeup(&pfs->pfs_lockowner); return rtval; } #if 0 /* * Get a string from userland into (buf). Strip a trailing * nl character (to allow easy access from the shell). * The buffer should be *buflenp + 1 chars long. vfs_getuserstr * will automatically add a nul char at the end. * * Returns 0 on success or the following errors * * EINVAL: file offset is non-zero. * EMSGSIZE: message is longer than kernel buffer * EFAULT: user i/o buffer is not addressable */ int vfs_getuserstr(uio, buf, buflenp) struct uio *uio; char *buf; int *buflenp; { int xlen; int error; if (uio->uio_offset != 0) return (EINVAL); xlen = *buflenp; /* must be able to read the whole string in one go */ if (xlen < uio->uio_resid) return (EMSGSIZE); xlen = uio->uio_resid; if ((error = uiomove(buf, xlen, uio)) != 0) return (error); /* allow multiple writes without seeks */ uio->uio_offset = 0; /* cleanup string and remove trailing newline */ buf[xlen] = '\0'; xlen = strlen(buf); if (xlen > 0 && buf[xlen-1] == '\n') buf[--xlen] = '\0'; *buflenp = xlen; return (0); } vfs_namemap_t * vfs_findname(nm, buf, buflen) vfs_namemap_t *nm; char *buf; int buflen; { for (; nm->nm_name; nm++) if (bcmp(buf, nm->nm_name, buflen+1) == 0) return (nm); return (0); } #endif void linprocfs_exit(struct proc *p) { struct pfsnode *pfs; pid_t pid = p->p_pid; /* * The reason for this loop is not obvious -- basicly, * linprocfs_freevp(), which is called via vgone() (eventually), * removes the specified procfs node from the pfshead list. * It does this by *pfsp = pfs->pfs_next, meaning that it * overwrites the node. So when we do pfs = pfs->next, we * end up skipping the node that replaces the one that was * vgone'd. Since it may have been the last one on the list, * it may also have been set to null -- but *our* pfs pointer, * here, doesn't see this. So the loop starts from the beginning * again. * * This is not a for() loop because the final event * would be "pfs = pfs->pfs_next"; in the case where * pfs is set to pfshead again, that would mean that * pfshead is skipped over. * */ pfs = pfshead; while (pfs) { if (pfs->pfs_pid == pid) { vgone(PFSTOV(pfs)); pfs = pfshead; } else pfs = pfs->pfs_next; } } Index: head/sys/i386/linux/linprocfs/linprocfs_vnops.c =================================================================== --- head/sys/i386/linux/linprocfs/linprocfs_vnops.c (revision 60859) +++ head/sys/i386/linux/linprocfs/linprocfs_vnops.c (revision 60860) @@ -1,994 +1,1023 @@ /* * Copyright (c) 1993, 1995 Jan-Simon Pendry * Copyright (c) 1993, 1995 * The Regents of the University of California. All rights reserved. * * This code is derived from software contributed to Berkeley by * Jan-Simon Pendry. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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. * 3. All advertising materials mentioning features or use of this software * must display the following acknowledgement: * This product includes software developed by the University of * California, Berkeley and its contributors. * 4. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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. * * @(#)procfs_vnops.c 8.18 (Berkeley) 5/21/95 * * $FreeBSD$ */ /* * procfs vnode interface */ #include #include #include #include #include #include #include #include #include #include #include #include +#include #include #include #include #include extern struct vnode *procfs_findtextvp __P((struct proc *)); extern int procfs_kmemaccess __P((struct proc *)); static int linprocfs_access __P((struct vop_access_args *)); static int linprocfs_badop __P((void)); static int linprocfs_bmap __P((struct vop_bmap_args *)); static int linprocfs_close __P((struct vop_close_args *)); static int linprocfs_getattr __P((struct vop_getattr_args *)); static int linprocfs_inactive __P((struct vop_inactive_args *)); static int linprocfs_ioctl __P((struct vop_ioctl_args *)); static int linprocfs_lookup __P((struct vop_lookup_args *)); static int linprocfs_open __P((struct vop_open_args *)); static int linprocfs_print __P((struct vop_print_args *)); static int linprocfs_readdir __P((struct vop_readdir_args *)); static int linprocfs_readlink __P((struct vop_readlink_args *)); static int linprocfs_reclaim __P((struct vop_reclaim_args *)); static int linprocfs_setattr __P((struct vop_setattr_args *)); /* * This is a list of the valid names in the * process-specific sub-directories. It is * used in linprocfs_lookup and linprocfs_readdir */ static struct proc_target { u_char pt_type; u_char pt_namlen; char *pt_name; pfstype pt_pfstype; int (*pt_valid) __P((struct proc *p)); } proc_targets[] = { #define N(s) sizeof(s)-1, s /* name type validp */ { DT_DIR, N("."), Pproc, NULL }, { DT_DIR, N(".."), Proot, NULL }, - { DT_REG, N("exe"), Pexe, linprocfs_validfile }, { DT_REG, N("mem"), Pmem, NULL }, + { DT_LNK, N("exe"), Pexe, NULL }, #undef N }; static const int nproc_targets = sizeof(proc_targets) / sizeof(proc_targets[0]); static pid_t atopid __P((const char *, u_int)); /* * set things up for doing i/o on * the pfsnode (vp). (vp) is locked * on entry, and should be left locked * on exit. * * for procfs we don't need to do anything * in particular for i/o. all that is done * is to support exclusive open on process * memory images. */ static int linprocfs_open(ap) struct vop_open_args /* { struct vnode *a_vp; int a_mode; struct ucred *a_cred; struct proc *a_p; } */ *ap; { struct pfsnode *pfs = VTOPFS(ap->a_vp); struct proc *p1, *p2; p2 = PFIND(pfs->pfs_pid); if (p2 == NULL) return (ENOENT); if (pfs->pfs_pid && !PRISON_CHECK(ap->a_p, p2)) return (ENOENT); switch (pfs->pfs_type) { case Pmem: if (((pfs->pfs_flags & FWRITE) && (ap->a_mode & O_EXCL)) || ((pfs->pfs_flags & O_EXCL) && (ap->a_mode & FWRITE))) return (EBUSY); p1 = ap->a_p; if (p_trespass(p1, p2) && !procfs_kmemaccess(p1)) return (EPERM); if (ap->a_mode & FWRITE) pfs->pfs_flags = ap->a_mode & (FWRITE|O_EXCL); return (0); default: break; } return (0); } /* * close the pfsnode (vp) after doing i/o. * (vp) is not locked on entry or exit. * * nothing to do for procfs other than undo * any exclusive open flag (see _open above). */ static int linprocfs_close(ap) struct vop_close_args /* { struct vnode *a_vp; int a_fflag; struct ucred *a_cred; struct proc *a_p; } */ *ap; { struct pfsnode *pfs = VTOPFS(ap->a_vp); struct proc *p; switch (pfs->pfs_type) { case Pmem: if ((ap->a_fflag & FWRITE) && (pfs->pfs_flags & O_EXCL)) pfs->pfs_flags &= ~(FWRITE|O_EXCL); /* * This rather complicated-looking code is trying to * determine if this was the last close on this particular * vnode. While one would expect v_usecount to be 1 at * that point, it seems that (according to John Dyson) * the VM system will bump up the usecount. So: if the * usecount is 2, and VOBJBUF is set, then this is really * the last close. Otherwise, if the usecount is < 2 * then it is definitely the last close. * If this is the last close, then it checks to see if * the target process has PF_LINGER set in p_pfsflags, * if this is *not* the case, then the process' stop flags * are cleared, and the process is woken up. This is * to help prevent the case where a process has been * told to stop on an event, but then the requesting process * has gone away or forgotten about it. */ if ((ap->a_vp->v_usecount < 2) && (p = pfind(pfs->pfs_pid)) && !(p->p_pfsflags & PF_LINGER)) { p->p_stops = 0; p->p_step = 0; wakeup(&p->p_step); } break; default: break; } return (0); } /* * do an ioctl operation on a pfsnode (vp). * (vp) is not locked on entry or exit. */ static int linprocfs_ioctl(ap) struct vop_ioctl_args *ap; { struct pfsnode *pfs = VTOPFS(ap->a_vp); struct proc *procp, *p; int error; int signo; struct procfs_status *psp; unsigned char flags; p = ap->a_p; procp = pfind(pfs->pfs_pid); if (procp == NULL) { return ENOTTY; } if (p_trespass(p, procp)) return EPERM; switch (ap->a_command) { case PIOCBIS: procp->p_stops |= *(unsigned int*)ap->a_data; break; case PIOCBIC: procp->p_stops &= ~*(unsigned int*)ap->a_data; break; case PIOCSFL: /* * NFLAGS is "non-suser_xxx flags" -- currently, only * PFS_ISUGID ("ignore set u/g id"); */ #define NFLAGS (PF_ISUGID) flags = (unsigned char)*(unsigned int*)ap->a_data; if (flags & NFLAGS && (error = suser(p))) return error; procp->p_pfsflags = flags; break; case PIOCGFL: *(unsigned int*)ap->a_data = (unsigned int)procp->p_pfsflags; case PIOCSTATUS: psp = (struct procfs_status *)ap->a_data; psp->state = (procp->p_step == 0); psp->flags = procp->p_pfsflags; psp->events = procp->p_stops; if (procp->p_step) { psp->why = procp->p_stype; psp->val = procp->p_xstat; } else { psp->why = psp->val = 0; /* Not defined values */ } break; case PIOCWAIT: psp = (struct procfs_status *)ap->a_data; if (procp->p_step == 0) { error = tsleep(&procp->p_stype, PWAIT | PCATCH, "piocwait", 0); if (error) return error; } psp->state = 1; /* It stopped */ psp->flags = procp->p_pfsflags; psp->events = procp->p_stops; psp->why = procp->p_stype; /* why it stopped */ psp->val = procp->p_xstat; /* any extra info */ break; case PIOCCONT: /* Restart a proc */ if (procp->p_step == 0) return EINVAL; /* Can only start a stopped process */ if ((signo = *(int*)ap->a_data) != 0) { if (signo >= NSIG || signo <= 0) return EINVAL; psignal(procp, signo); } procp->p_step = 0; wakeup(&procp->p_step); break; default: return (ENOTTY); } return 0; } /* * do block mapping for pfsnode (vp). * since we don't use the buffer cache * for procfs this function should never * be called. in any case, it's not clear * what part of the kernel ever makes use * of this function. for sanity, this is the * usual no-op bmap, although returning * (EIO) would be a reasonable alternative. */ static int linprocfs_bmap(ap) struct vop_bmap_args /* { struct vnode *a_vp; daddr_t a_bn; struct vnode **a_vpp; daddr_t *a_bnp; int *a_runp; } */ *ap; { if (ap->a_vpp != NULL) *ap->a_vpp = ap->a_vp; if (ap->a_bnp != NULL) *ap->a_bnp = ap->a_bn; if (ap->a_runp != NULL) *ap->a_runp = 0; return (0); } /* * linprocfs_inactive is called when the pfsnode * is vrele'd and the reference count goes * to zero. (vp) will be on the vnode free * list, so to get it back vget() must be * used. * * (vp) is locked on entry, but must be unlocked on exit. */ static int linprocfs_inactive(ap) struct vop_inactive_args /* { struct vnode *a_vp; } */ *ap; { struct vnode *vp = ap->a_vp; VOP_UNLOCK(vp, 0, ap->a_p); return (0); } /* * _reclaim is called when getnewvnode() * wants to make use of an entry on the vnode * free list. at this time the filesystem needs * to free any private data and remove the node * from any private lists. */ static int linprocfs_reclaim(ap) struct vop_reclaim_args /* { struct vnode *a_vp; } */ *ap; { return (linprocfs_freevp(ap->a_vp)); } /* * _print is used for debugging. * just print a readable description * of (vp). */ static int linprocfs_print(ap) struct vop_print_args /* { struct vnode *a_vp; } */ *ap; { struct pfsnode *pfs = VTOPFS(ap->a_vp); printf("tag VT_PROCFS, type %d, pid %ld, mode %x, flags %lx\n", pfs->pfs_type, (long)pfs->pfs_pid, pfs->pfs_mode, pfs->pfs_flags); return (0); } /* * generic entry point for unsupported operations */ static int linprocfs_badop() { return (EIO); } /* * Invent attributes for pfsnode (vp) and store * them in (vap). * Directories lengths are returned as zero since * any real length would require the genuine size * to be computed, and nothing cares anyway. * * this is relatively minimal for procfs. */ static int linprocfs_getattr(ap) struct vop_getattr_args /* { struct vnode *a_vp; struct vattr *a_vap; struct ucred *a_cred; struct proc *a_p; } */ *ap; { struct pfsnode *pfs = VTOPFS(ap->a_vp); struct vattr *vap = ap->a_vap; struct proc *procp; int error; /* * First make sure that the process and its credentials * still exist. */ switch (pfs->pfs_type) { case Proot: case Pself: procp = 0; break; default: procp = PFIND(pfs->pfs_pid); if (procp == 0 || procp->p_cred == NULL || procp->p_ucred == NULL) return (ENOENT); } error = 0; /* start by zeroing out the attributes */ VATTR_NULL(vap); /* next do all the common fields */ vap->va_type = ap->a_vp->v_type; vap->va_mode = pfs->pfs_mode; vap->va_fileid = pfs->pfs_fileno; vap->va_flags = 0; vap->va_blocksize = PAGE_SIZE; vap->va_bytes = vap->va_size = 0; vap->va_fsid = ap->a_vp->v_mount->mnt_stat.f_fsid.val[0]; /* * Make all times be current TOD. * It would be possible to get the process start * time from the p_stat structure, but there's * no "file creation" time stamp anyway, and the * p_stat structure is not addressible if u. gets * swapped out for that process. */ nanotime(&vap->va_ctime); vap->va_atime = vap->va_mtime = vap->va_ctime; /* * If the process has exercised some setuid or setgid * privilege, then rip away read/write permission so * that only root can gain access. */ switch (pfs->pfs_type) { case Pmem: /* Retain group kmem readablity. */ if (procp->p_flag & P_SUGID) vap->va_mode &= ~(VREAD|VWRITE); break; default: break; } /* * now do the object specific fields * * The size could be set from struct reg, but it's hardly * worth the trouble, and it puts some (potentially) machine * dependent data into this machine-independent code. If it * becomes important then this function should break out into * a per-file stat function in the corresponding .c file. */ vap->va_nlink = 1; if (procp) { vap->va_uid = procp->p_ucred->cr_uid; vap->va_gid = procp->p_ucred->cr_gid; } switch (pfs->pfs_type) { case Proot: /* * Set nlink to 1 to tell fts(3) we don't actually know. */ vap->va_nlink = 1; vap->va_uid = 0; vap->va_gid = 0; vap->va_size = vap->va_bytes = DEV_BSIZE; break; case Pself: { char buf[16]; /* should be enough */ vap->va_uid = 0; vap->va_gid = 0; vap->va_size = vap->va_bytes = snprintf(buf, sizeof(buf), "%ld", (long)curproc->p_pid); break; } case Pproc: vap->va_nlink = nproc_targets; vap->va_size = vap->va_bytes = DEV_BSIZE; break; - case Pexe: - error = EOPNOTSUPP; + case Pexe: { + char *fullpath, *freepath; + error = textvp_fullpath(procp, &fullpath, &freepath); + if (error == 0) { + vap->va_size = strlen(fullpath); + free(freepath, M_TEMP); + } else { + vap->va_size = sizeof("unknown") - 1; + error = 0; + } + vap->va_bytes = vap->va_size; break; + } case Pmeminfo: case Pcpuinfo: break; case Pmem: /* * If we denied owner access earlier, then we have to * change the owner to root - otherwise 'ps' and friends * will break even though they are setgid kmem. *SIGH* */ if (procp->p_flag & P_SUGID) vap->va_uid = 0; else vap->va_uid = procp->p_ucred->cr_uid; vap->va_gid = KMEM_GROUP; break; default: panic("linprocfs_getattr"); } return (error); } static int linprocfs_setattr(ap) struct vop_setattr_args /* { struct vnode *a_vp; struct vattr *a_vap; struct ucred *a_cred; struct proc *a_p; } */ *ap; { if (ap->a_vap->va_flags != VNOVAL) return (EOPNOTSUPP); /* * just fake out attribute setting * it's not good to generate an error * return, otherwise things like creat() * will fail when they try to set the * file length to 0. worse, this means * that echo $note > /proc/$pid/note will fail. */ return (0); } /* * implement access checking. * * something very similar to this code is duplicated * throughout the 4bsd kernel and should be moved * into kern/vfs_subr.c sometime. * * actually, the check for super-user is slightly * broken since it will allow read access to write-only * objects. this doesn't cause any particular trouble * but does mean that the i/o entry points need to check * that the operation really does make sense. */ static int linprocfs_access(ap) struct vop_access_args /* { struct vnode *a_vp; int a_mode; struct ucred *a_cred; struct proc *a_p; } */ *ap; { struct vattr *vap; struct vattr vattr; int error; /* * If you're the super-user, * you always get access. */ if (ap->a_cred->cr_uid == 0) return (0); vap = &vattr; error = VOP_GETATTR(ap->a_vp, vap, ap->a_cred, ap->a_p); if (error) return (error); /* * Access check is based on only one of owner, group, public. * If not owner, then check group. If not a member of the * group, then check public access. */ if (ap->a_cred->cr_uid != vap->va_uid) { gid_t *gp; int i; ap->a_mode >>= 3; gp = ap->a_cred->cr_groups; for (i = 0; i < ap->a_cred->cr_ngroups; i++, gp++) if (vap->va_gid == *gp) goto found; ap->a_mode >>= 3; found: ; } if ((vap->va_mode & ap->a_mode) == ap->a_mode) return (0); return (EACCES); } /* * lookup. this is incredibly complicated in the * general case, however for most pseudo-filesystems * very little needs to be done. * * unless you want to get a migraine, just make sure your * filesystem doesn't do any locking of its own. otherwise * read and inwardly digest ufs_lookup(). */ static int linprocfs_lookup(ap) struct vop_lookup_args /* { struct vnode * a_dvp; struct vnode ** a_vpp; struct componentname * a_cnp; } */ *ap; { struct componentname *cnp = ap->a_cnp; struct vnode **vpp = ap->a_vpp; struct vnode *dvp = ap->a_dvp; char *pname = cnp->cn_nameptr; - struct proc *curp = cnp->cn_proc; struct proc_target *pt; - struct vnode *fvp; pid_t pid; struct pfsnode *pfs; struct proc *p; int i; *vpp = NULL; if (cnp->cn_nameiop == DELETE || cnp->cn_nameiop == RENAME) return (EROFS); if (cnp->cn_namelen == 1 && *pname == '.') { *vpp = dvp; VREF(dvp); /* vn_lock(dvp, LK_EXCLUSIVE | LK_RETRY, curp); */ return (0); } pfs = VTOPFS(dvp); switch (pfs->pfs_type) { case Proot: if (cnp->cn_flags & ISDOTDOT) return (EIO); if (CNEQ(cnp, "self", 4)) return (linprocfs_allocvp(dvp->v_mount, vpp, 0, Pself)); if (CNEQ(cnp, "meminfo", 7)) return (linprocfs_allocvp(dvp->v_mount, vpp, 0, Pmeminfo)); if (CNEQ(cnp, "cpuinfo", 7)) return (linprocfs_allocvp(dvp->v_mount, vpp, 0, Pcpuinfo)); pid = atopid(pname, cnp->cn_namelen); if (pid == NO_PID) break; p = PFIND(pid); if (p == 0) break; return (linprocfs_allocvp(dvp->v_mount, vpp, pid, Pproc)); case Pproc: if (cnp->cn_flags & ISDOTDOT) return (linprocfs_root(dvp->v_mount, vpp)); p = PFIND(pfs->pfs_pid); if (p == 0) break; for (pt = proc_targets, i = 0; i < nproc_targets; pt++, i++) { if (cnp->cn_namelen == pt->pt_namlen && bcmp(pt->pt_name, pname, cnp->cn_namelen) == 0 && (pt->pt_valid == NULL || (*pt->pt_valid)(p))) goto found; } break; found: - if (pt->pt_pfstype == Pexe) { - fvp = procfs_findtextvp(p); - /* We already checked that it exists. */ - VREF(fvp); - vn_lock(fvp, LK_EXCLUSIVE | LK_RETRY, curp); - *vpp = fvp; - return (0); - } - return (linprocfs_allocvp(dvp->v_mount, vpp, pfs->pfs_pid, pt->pt_pfstype)); default: return (ENOTDIR); } return (cnp->cn_nameiop == LOOKUP ? ENOENT : EROFS); } /* * Does this process have a text file? */ int linprocfs_validfile(p) struct proc *p; { return (procfs_findtextvp(p) != NULLVP); } /* * readdir() returns directory entries from pfsnode (vp). * * We generate just one directory entry at a time, as it would probably * not pay off to buffer several entries locally to save uiomove calls. */ static int linprocfs_readdir(ap) struct vop_readdir_args /* { struct vnode *a_vp; struct uio *a_uio; struct ucred *a_cred; int *a_eofflag; int *a_ncookies; u_long **a_cookies; } */ *ap; { struct uio *uio = ap->a_uio; struct dirent d; struct dirent *dp = &d; struct pfsnode *pfs; int count, error, i, off; static u_int delen; if (!delen) { d.d_namlen = PROCFS_NAMELEN; delen = GENERIC_DIRSIZ(&d); } pfs = VTOPFS(ap->a_vp); off = (int)uio->uio_offset; if (off != uio->uio_offset || off < 0 || off % delen != 0 || uio->uio_resid < delen) return (EINVAL); error = 0; count = 0; i = off / delen; switch (pfs->pfs_type) { /* * this is for the process-specific sub-directories. * all that is needed to is copy out all the entries * from the procent[] table (top of this file). */ case Pproc: { struct proc *p; struct proc_target *pt; p = PFIND(pfs->pfs_pid); if (p == NULL) break; if (!PRISON_CHECK(curproc, p)) break; for (pt = &proc_targets[i]; uio->uio_resid >= delen && i < nproc_targets; pt++, i++) { if (pt->pt_valid && (*pt->pt_valid)(p) == 0) continue; dp->d_reclen = delen; dp->d_fileno = PROCFS_FILENO(pfs->pfs_pid, pt->pt_pfstype); dp->d_namlen = pt->pt_namlen; bcopy(pt->pt_name, dp->d_name, pt->pt_namlen + 1); dp->d_type = pt->pt_type; if ((error = uiomove((caddr_t)dp, delen, uio)) != 0) break; } break; } /* * this is for the root of the procfs filesystem * what is needed is a special entry for "self" * followed by an entry for each process on allproc #ifdef PROCFS_ZOMBIE * and zombproc. #endif */ case Proot: { #ifdef PROCFS_ZOMBIE int doingzomb = 0; #endif int pcnt = 0; volatile struct proc *p = allproc.lh_first; for (; p && uio->uio_resid >= delen; i++, pcnt++) { bzero((char *) dp, delen); dp->d_reclen = delen; switch (i) { case 0: /* `.' */ case 1: /* `..' */ dp->d_fileno = PROCFS_FILENO(0, Proot); dp->d_namlen = i + 1; bcopy("..", dp->d_name, dp->d_namlen); dp->d_name[i + 1] = '\0'; dp->d_type = DT_DIR; break; case 2: dp->d_fileno = PROCFS_FILENO(0, Pself); dp->d_namlen = 4; bcopy("self", dp->d_name, 5); dp->d_type = DT_LNK; break; case 3: dp->d_fileno = PROCFS_FILENO(0, Pmeminfo); dp->d_namlen = 7; bcopy("meminfo", dp->d_name, 8); dp->d_type = DT_REG; break; case 4: dp->d_fileno = PROCFS_FILENO(0, Pcpuinfo); dp->d_namlen = 7; bcopy("cpuinfo", dp->d_name, 8); dp->d_type = DT_REG; break; default: while (pcnt < i) { p = p->p_list.le_next; if (!p) goto done; if (!PRISON_CHECK(curproc, p)) continue; pcnt++; } while (!PRISON_CHECK(curproc, p)) { p = p->p_list.le_next; if (!p) goto done; } dp->d_fileno = PROCFS_FILENO(p->p_pid, Pproc); dp->d_namlen = sprintf(dp->d_name, "%ld", (long)p->p_pid); dp->d_type = DT_REG; p = p->p_list.le_next; break; } if ((error = uiomove((caddr_t)dp, delen, uio)) != 0) break; } done: #ifdef PROCFS_ZOMBIE if (p == 0 && doingzomb == 0) { doingzomb = 1; p = zombproc.lh_first; goto again; } #endif break; } default: error = ENOTDIR; break; } uio->uio_offset = i * delen; return (error); } /* - * readlink reads the link of `self' + * readlink reads the link of `self' or `exe' */ static int linprocfs_readlink(ap) struct vop_readlink_args *ap; { char buf[16]; /* should be enough */ - int len; + struct proc *procp; + struct vnode *vp = ap->a_vp; + struct pfsnode *pfs = VTOPFS(vp); + char *fullpath, *freepath; + int error, len; - if (VTOPFS(ap->a_vp)->pfs_fileno != PROCFS_FILENO(0, Pself)) - return (EINVAL); + switch (pfs->pfs_type) { + case Pself: + if (pfs->pfs_fileno != PROCFS_FILENO(0, Pself)) + return (EINVAL); - len = snprintf(buf, sizeof(buf), "%ld", (long)curproc->p_pid); + len = snprintf(buf, sizeof(buf), "%ld", (long)curproc->p_pid); - return (uiomove((caddr_t)buf, len, ap->a_uio)); + return (uiomove(buf, len, ap->a_uio)); + /* + * There _should_ be no way for an entire process to disappear + * from under us... + */ + case Pexe: + procp = PFIND(pfs->pfs_pid); + if (procp == NULL || procp->p_cred == NULL || + procp->p_ucred == NULL) { + printf("linprocfs_readlink: pid %d disappeared\n", + pfs->pfs_pid); + return (uiomove("unknown", sizeof("unknown") - 1, + ap->a_uio)); + } + error = textvp_fullpath(procp, &fullpath, &freepath); + if (error != 0) + return (uiomove("unknown", sizeof("unknown") - 1, + ap->a_uio)); + error = uiomove(fullpath, strlen(fullpath), ap->a_uio); + free(freepath, M_TEMP); + return (error); + default: + return (EINVAL); + } } /* * convert decimal ascii to pid_t */ static pid_t atopid(b, len) const char *b; u_int len; { pid_t p = 0; while (len--) { char c = *b++; if (c < '0' || c > '9') return (NO_PID); p = 10 * p + (c - '0'); if (p > PID_MAX) return (NO_PID); } return (p); } /* * procfs vnode operations. */ vop_t **linprocfs_vnodeop_p; static struct vnodeopv_entry_desc linprocfs_vnodeop_entries[] = { { &vop_default_desc, (vop_t *) vop_defaultop }, { &vop_access_desc, (vop_t *) linprocfs_access }, { &vop_advlock_desc, (vop_t *) linprocfs_badop }, { &vop_bmap_desc, (vop_t *) linprocfs_bmap }, { &vop_close_desc, (vop_t *) linprocfs_close }, { &vop_create_desc, (vop_t *) linprocfs_badop }, { &vop_getattr_desc, (vop_t *) linprocfs_getattr }, { &vop_inactive_desc, (vop_t *) linprocfs_inactive }, { &vop_link_desc, (vop_t *) linprocfs_badop }, { &vop_lookup_desc, (vop_t *) linprocfs_lookup }, { &vop_mkdir_desc, (vop_t *) linprocfs_badop }, { &vop_mknod_desc, (vop_t *) linprocfs_badop }, { &vop_open_desc, (vop_t *) linprocfs_open }, { &vop_pathconf_desc, (vop_t *) vop_stdpathconf }, { &vop_print_desc, (vop_t *) linprocfs_print }, { &vop_read_desc, (vop_t *) linprocfs_rw }, { &vop_readdir_desc, (vop_t *) linprocfs_readdir }, { &vop_readlink_desc, (vop_t *) linprocfs_readlink }, { &vop_reclaim_desc, (vop_t *) linprocfs_reclaim }, { &vop_remove_desc, (vop_t *) linprocfs_badop }, { &vop_rename_desc, (vop_t *) linprocfs_badop }, { &vop_rmdir_desc, (vop_t *) linprocfs_badop }, { &vop_setattr_desc, (vop_t *) linprocfs_setattr }, { &vop_symlink_desc, (vop_t *) linprocfs_badop }, { &vop_write_desc, (vop_t *) linprocfs_rw }, { &vop_ioctl_desc, (vop_t *) linprocfs_ioctl }, { NULL, NULL } }; static struct vnodeopv_desc linprocfs_vnodeop_opv_desc = { &linprocfs_vnodeop_p, linprocfs_vnodeop_entries }; VNODEOP_SET(linprocfs_vnodeop_opv_desc);