aboutsummaryrefslogtreecommitdiff
path: root/sys/kern/uipc_shm.c
diff options
context:
space:
mode:
Diffstat (limited to 'sys/kern/uipc_shm.c')
-rw-r--r--sys/kern/uipc_shm.c76
1 files changed, 51 insertions, 25 deletions
diff --git a/sys/kern/uipc_shm.c b/sys/kern/uipc_shm.c
index b4016e9dd6bf..85fe48ddd466 100644
--- a/sys/kern/uipc_shm.c
+++ b/sys/kern/uipc_shm.c
@@ -100,6 +100,7 @@
#include <vm/vm_page.h>
#include <vm/vm_pageout.h>
#include <vm/vm_pager.h>
+#include <vm/vm_radix.h>
#include <vm/swap_pager.h>
struct shm_mapping {
@@ -195,6 +196,7 @@ SYSCTL_INT(_vm_largepages, OID_AUTO, reclaim_tries,
static int
uiomove_object_page(vm_object_t obj, size_t len, struct uio *uio)
{
+ struct pctrie_iter pages;
vm_page_t m;
vm_pindex_t idx;
size_t tlen;
@@ -214,8 +216,9 @@ uiomove_object_page(vm_object_t obj, size_t len, struct uio *uio)
* page: use zero_region. This is intended to avoid instantiating
* pages on read from a sparse region.
*/
+ vm_page_iter_init(&pages, obj);
VM_OBJECT_WLOCK(obj);
- m = vm_page_lookup(obj, idx);
+ m = vm_radix_iter_lookup(&pages, idx);
if (uio->uio_rw == UIO_READ && m == NULL &&
!vm_pager_has_page(obj, idx, NULL, NULL)) {
VM_OBJECT_WUNLOCK(obj);
@@ -229,8 +232,8 @@ uiomove_object_page(vm_object_t obj, size_t len, struct uio *uio)
* lock to page out tobj's pages because tobj is a OBJT_SWAP
* type object.
*/
- rv = vm_page_grab_valid(&m, obj, idx,
- VM_ALLOC_NORMAL | VM_ALLOC_SBUSY | VM_ALLOC_IGN_SBUSY);
+ rv = vm_page_grab_valid_iter(&m, obj, idx,
+ VM_ALLOC_NORMAL | VM_ALLOC_SBUSY | VM_ALLOC_IGN_SBUSY, &pages);
if (rv != VM_PAGER_OK) {
VM_OBJECT_WUNLOCK(obj);
if (bootverbose) {
@@ -481,7 +484,10 @@ shm_write(struct file *fp, struct uio *uio, struct ucred *active_cred,
struct shmfd *shmfd;
void *rl_cookie;
int error;
- off_t size;
+ off_t newsize;
+
+ KASSERT((flags & FOF_OFFSET) == 0 || uio->uio_offset >= 0,
+ ("%s: negative offset", __func__));
shmfd = fp->f_data;
#ifdef MAC
@@ -503,21 +509,23 @@ shm_write(struct file *fp, struct uio *uio, struct ucred *active_cred,
return (EFBIG);
}
- size = shmfd->shm_size;
+ newsize = atomic_load_64(&shmfd->shm_size);
} else {
- size = uio->uio_offset + uio->uio_resid;
+ newsize = uio->uio_offset + uio->uio_resid;
}
if ((flags & FOF_OFFSET) == 0)
rl_cookie = shm_rangelock_wlock(shmfd, 0, OFF_MAX);
else
- rl_cookie = shm_rangelock_wlock(shmfd, uio->uio_offset, size);
+ rl_cookie = shm_rangelock_wlock(shmfd, uio->uio_offset,
+ MAX(newsize, uio->uio_offset));
if ((shmfd->shm_seals & F_SEAL_WRITE) != 0) {
error = EPERM;
} else {
error = 0;
if ((shmfd->shm_flags & SHM_GROW_ON_WRITE) != 0 &&
- size > shmfd->shm_size) {
- error = shm_dotruncate_cookie(shmfd, size, rl_cookie);
+ newsize > shmfd->shm_size) {
+ error = shm_dotruncate_cookie(shmfd, newsize,
+ rl_cookie);
}
if (error == 0)
error = uiomove_object(shmfd->shm_object,
@@ -1126,10 +1134,10 @@ shm_doremove(struct shm_mapping *map)
int
kern_shm_open2(struct thread *td, const char *userpath, int flags, mode_t mode,
- int shmflags, struct filecaps *fcaps, const char *name __unused)
+ int shmflags, struct filecaps *fcaps, const char *name __unused,
+ struct shmfd *shmfd)
{
struct pwddesc *pdp;
- struct shmfd *shmfd;
struct file *fp;
char *path;
void *rl_cookie;
@@ -1206,23 +1214,41 @@ kern_shm_open2(struct thread *td, const char *userpath, int flags, mode_t mode,
if (error != 0)
goto outnofp;
- /* A SHM_ANON path pointer creates an anonymous object. */
+ /*
+ * A SHM_ANON path pointer creates an anonymous object. We allow other
+ * parts of the kernel to pre-populate a shmfd and then materialize an
+ * fd for it here as a means to pass data back up to userland. This
+ * doesn't really make sense for named shm objects, but it makes plenty
+ * of sense for anonymous objects.
+ */
if (userpath == SHM_ANON) {
- /* A read-only anonymous object is pointless. */
- if ((flags & O_ACCMODE) == O_RDONLY) {
- error = EINVAL;
- goto out;
- }
- shmfd = shm_alloc(td->td_ucred, cmode, largepage);
- if (shmfd == NULL) {
- error = ENOMEM;
- goto out;
+ if (shmfd != NULL) {
+ shm_hold(shmfd);
+ } else {
+ /*
+ * A read-only anonymous object is pointless, unless it
+ * was pre-populated by the kernel with the expectation
+ * that a shmfd would later be created for userland to
+ * access it through.
+ */
+ if ((flags & O_ACCMODE) == O_RDONLY) {
+ error = EINVAL;
+ goto out;
+ }
+ shmfd = shm_alloc(td->td_ucred, cmode, largepage);
+ if (shmfd == NULL) {
+ error = ENOMEM;
+ goto out;
+ }
+
+ shmfd->shm_seals = initial_seals;
+ shmfd->shm_flags = shmflags;
}
- shmfd->shm_seals = initial_seals;
- shmfd->shm_flags = shmflags;
} else {
fnv = fnv_32_str(path, FNV1_32_INIT);
sx_xlock(&shm_dict_lock);
+
+ MPASS(shmfd == NULL);
shmfd = shm_lookup(path, fnv);
if (shmfd == NULL) {
/* Object does not yet exist, create it if requested. */
@@ -2165,7 +2191,7 @@ kern_shm_open(struct thread *td, const char *path, int flags, mode_t mode,
struct filecaps *caps)
{
- return (kern_shm_open2(td, path, flags, mode, 0, caps, NULL));
+ return (kern_shm_open2(td, path, flags, mode, 0, caps, NULL, NULL));
}
/*
@@ -2183,7 +2209,7 @@ sys_shm_open2(struct thread *td, struct shm_open2_args *uap)
{
return (kern_shm_open2(td, uap->path, uap->flags, uap->mode,
- uap->shmflags, NULL, uap->name));
+ uap->shmflags, NULL, uap->name, NULL));
}
int
generated by cgit v1.2.3 (git 2.25.1) at 2025-08-11 12:14:09 +0000