| Message ID | 20180507083807.28792-24-mszeredi@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 05/07/2018 01:37 AM, Miklos Szeredi wrote: > A corner case of a corner case is when > > - file opened for O_RDONLY > - which is then memory mapped SHARED > - file opened for O_WRONLY > - contents modified > - contents read back though the shared mapping > > Unfortunately it looks very difficult to do anything about the established > shared map after the file is copied up. > > Instead, when a read-only file is mapped shared, copy up the file before > actually doing the map. This may result in unnecessary copy-ups (but so > may copy-up on open(O_RDWR) for exampe). for example). > > We can revisit this later if it turns out to be a performance problem in > real life. > > Signed-off-by: Miklos Szeredi <mszeredi@redhat.com> > --- > fs/overlayfs/Kconfig | 21 +++++++++++++++++++++ > fs/overlayfs/file.c | 22 ++++++++++++++++++++++ > fs/overlayfs/overlayfs.h | 7 +++++++ > fs/overlayfs/ovl_entry.h | 1 + > fs/overlayfs/super.c | 22 ++++++++++++++++++++++ > 5 files changed, 73 insertions(+) > > diff --git a/fs/overlayfs/Kconfig b/fs/overlayfs/Kconfig > index 17032631c5cf..991c0a5a0e00 100644 > --- a/fs/overlayfs/Kconfig > +++ b/fs/overlayfs/Kconfig > @@ -103,3 +103,24 @@ config OVERLAY_FS_XINO_AUTO > For more information, see Documentation/filesystems/overlayfs.txt > > If unsure, say N. > + > +config OVERLAY_FS_COPY_UP_SHARED > + bool "Overlayfs: copy up when mapping a file shared" a shared file" ?? > + default n > + depends on OVERLAY_FS > + help > + If this option is enabled then on mapping a file with MAP_SHARED > + overlayfs copies up the file in anticipation of it being modified (just > + like we copy up the file on O_WRONLY and O_RDWR in anticipation of > + modification). This does not interfere with shared library loading, as > + that uses MAP_PRIVATE. But there might be use cases out there where > + this impacts performance and disk usage. > + > + This just selects the default, the feature can also be enabled or > + disabled in the running kernel or individually on each overlay mount. > + > + To get maximally standard compliant behavior, enable this option. > + > + To get a maximally backward compatible kernel, disable this option. > + > + If unsure, say N. For Kconfig (coding-style.rst): Lines under a ``config`` definition are indented with one tab, while help text is indented an additional two spaces.
On Mon, May 7, 2018 at 9:28 PM, Randy Dunlap <rdunlap@infradead.org> wrote: > On 05/07/2018 01:37 AM, Miklos Szeredi wrote: >> A corner case of a corner case is when >> >> - file opened for O_RDONLY >> - which is then memory mapped SHARED >> - file opened for O_WRONLY >> - contents modified >> - contents read back though the shared mapping >> >> Unfortunately it looks very difficult to do anything about the established >> shared map after the file is copied up. >> >> Instead, when a read-only file is mapped shared, copy up the file before >> actually doing the map. This may result in unnecessary copy-ups (but so >> may copy-up on open(O_RDWR) for exampe). > > for example). > >> >> We can revisit this later if it turns out to be a performance problem in >> real life. >> >> Signed-off-by: Miklos Szeredi <mszeredi@redhat.com> >> --- >> fs/overlayfs/Kconfig | 21 +++++++++++++++++++++ >> fs/overlayfs/file.c | 22 ++++++++++++++++++++++ >> fs/overlayfs/overlayfs.h | 7 +++++++ >> fs/overlayfs/ovl_entry.h | 1 + >> fs/overlayfs/super.c | 22 ++++++++++++++++++++++ >> 5 files changed, 73 insertions(+) >> >> diff --git a/fs/overlayfs/Kconfig b/fs/overlayfs/Kconfig >> index 17032631c5cf..991c0a5a0e00 100644 >> --- a/fs/overlayfs/Kconfig >> +++ b/fs/overlayfs/Kconfig >> @@ -103,3 +103,24 @@ config OVERLAY_FS_XINO_AUTO >> For more information, see Documentation/filesystems/overlayfs.txt >> >> If unsure, say N. >> + >> +config OVERLAY_FS_COPY_UP_SHARED >> + bool "Overlayfs: copy up when mapping a file shared" > > a shared file" ?? This is referring to MAP_SHARED flag of mmap(). So it's the mapping that is shared, not the file. > >> + default n >> + depends on OVERLAY_FS >> + help >> + If this option is enabled then on mapping a file with MAP_SHARED >> + overlayfs copies up the file in anticipation of it being modified (just >> + like we copy up the file on O_WRONLY and O_RDWR in anticipation of >> + modification). This does not interfere with shared library loading, as >> + that uses MAP_PRIVATE. But there might be use cases out there where >> + this impacts performance and disk usage. >> + >> + This just selects the default, the feature can also be enabled or >> + disabled in the running kernel or individually on each overlay mount. >> + >> + To get maximally standard compliant behavior, enable this option. >> + >> + To get a maximally backward compatible kernel, disable this option. >> + >> + If unsure, say N. > > For Kconfig (coding-style.rst): > Lines under a ``config`` definition are indented with one tab, while help text > is indented an additional two spaces. Not sure what went wrong there. Fixed now. Thanks, Miklos
diff --git a/fs/overlayfs/Kconfig b/fs/overlayfs/Kconfig index 17032631c5cf..991c0a5a0e00 100644 --- a/fs/overlayfs/Kconfig +++ b/fs/overlayfs/Kconfig @@ -103,3 +103,24 @@ config OVERLAY_FS_XINO_AUTO For more information, see Documentation/filesystems/overlayfs.txt If unsure, say N. + +config OVERLAY_FS_COPY_UP_SHARED + bool "Overlayfs: copy up when mapping a file shared" + default n + depends on OVERLAY_FS + help + If this option is enabled then on mapping a file with MAP_SHARED + overlayfs copies up the file in anticipation of it being modified (just + like we copy up the file on O_WRONLY and O_RDWR in anticipation of + modification). This does not interfere with shared library loading, as + that uses MAP_PRIVATE. But there might be use cases out there where + this impacts performance and disk usage. + + This just selects the default, the feature can also be enabled or + disabled in the running kernel or individually on each overlay mount. + + To get maximally standard compliant behavior, enable this option. + + To get a maximally backward compatible kernel, disable this option. + + If unsure, say N. diff --git a/fs/overlayfs/file.c b/fs/overlayfs/file.c index 2ac95c95e8e6..a60734ec89ec 100644 --- a/fs/overlayfs/file.c +++ b/fs/overlayfs/file.c @@ -10,6 +10,7 @@ #include <linux/file.h> #include <linux/mount.h> #include <linux/xattr.h> +#include <linux/mman.h> #include <linux/uio.h> #include "overlayfs.h" @@ -256,6 +257,26 @@ static int ovl_fsync(struct file *file, loff_t start, loff_t end, int datasync) return ret; } +static int ovl_pre_mmap(struct file *file, unsigned long prot, + unsigned long flag) +{ + int err = 0; + + /* + * Take MAP_SHARED as hint about future writes to the file (through + * another file descriptor). Caller might not have had such an intent, + * but we hope MAP_PRIVATE will be used in most such cases. + * + * If we don't copy up now and the file is modified, it becomes really + * difficult to change the mapping to match that of the file's content + * later. + */ + if ((flag & MAP_SHARED) && ovl_copy_up_shared(file_inode(file)->i_sb)) + err = ovl_copy_up(file_dentry(file)); + + return err; +} + static int ovl_mmap(struct file *file, struct vm_area_struct *vma) { struct fd real; @@ -473,6 +494,7 @@ const struct file_operations ovl_file_operations = { .read_iter = ovl_read_iter, .write_iter = ovl_write_iter, .fsync = ovl_fsync, + .pre_mmap = ovl_pre_mmap, .mmap = ovl_mmap, .fallocate = ovl_fallocate, .unlocked_ioctl = ovl_ioctl, diff --git a/fs/overlayfs/overlayfs.h b/fs/overlayfs/overlayfs.h index edbf69c8f45d..caaa47cea2aa 100644 --- a/fs/overlayfs/overlayfs.h +++ b/fs/overlayfs/overlayfs.h @@ -272,6 +272,13 @@ static inline unsigned int ovl_xino_bits(struct super_block *sb) return ofs->xino_bits; } +static inline bool ovl_copy_up_shared(struct super_block *sb) +{ + struct ovl_fs *ofs = sb->s_fs_info; + + return !(sb->s_flags & SB_RDONLY) && ofs->config.copy_up_shared; +} + /* namei.c */ int ovl_check_fh_len(struct ovl_fh *fh, int fh_len); diff --git a/fs/overlayfs/ovl_entry.h b/fs/overlayfs/ovl_entry.h index 41655a7d6894..3bea47c63fd9 100644 --- a/fs/overlayfs/ovl_entry.h +++ b/fs/overlayfs/ovl_entry.h @@ -18,6 +18,7 @@ struct ovl_config { const char *redirect_mode; bool index; bool nfs_export; + bool copy_up_shared; int xino; }; diff --git a/fs/overlayfs/super.c b/fs/overlayfs/super.c index ad6a5baf226b..c3d8c7ea180f 100644 --- a/fs/overlayfs/super.c +++ b/fs/overlayfs/super.c @@ -56,6 +56,12 @@ module_param_named(xino_auto, ovl_xino_auto_def, bool, 0644); MODULE_PARM_DESC(ovl_xino_auto_def, "Auto enable xino feature"); +static bool ovl_copy_up_shared_def = + IS_ENABLED(CONFIG_OVERLAY_FS_COPY_UP_SHARED); +module_param_named(copy_up_shared, ovl_copy_up_shared_def, bool, 0644); +MODULE_PARM_DESC(ovl_copy_up_shared_def, + "Copy up when mapping a file shared"); + static void ovl_entry_stack_free(struct ovl_entry *oe) { unsigned int i; @@ -380,6 +386,9 @@ static int ovl_show_options(struct seq_file *m, struct dentry *dentry) "on" : "off"); if (ofs->config.xino != ovl_xino_def()) seq_printf(m, ",xino=%s", ovl_xino_str[ofs->config.xino]); + if (ofs->config.copy_up_shared != ovl_copy_up_shared_def) + seq_printf(m, ",copy_up_shared=%s", + ofs->config.copy_up_shared ? "on" : "off"); return 0; } @@ -417,6 +426,8 @@ enum { OPT_XINO_ON, OPT_XINO_OFF, OPT_XINO_AUTO, + OPT_COPY_UP_SHARED_ON, + OPT_COPY_UP_SHARED_OFF, OPT_ERR, }; @@ -433,6 +444,8 @@ static const match_table_t ovl_tokens = { {OPT_XINO_ON, "xino=on"}, {OPT_XINO_OFF, "xino=off"}, {OPT_XINO_AUTO, "xino=auto"}, + {OPT_COPY_UP_SHARED_ON, "copy_up_shared=on"}, + {OPT_COPY_UP_SHARED_OFF, "copy_up_shared=off"}, {OPT_ERR, NULL} }; @@ -559,6 +572,14 @@ static int ovl_parse_opt(char *opt, struct ovl_config *config) config->xino = OVL_XINO_AUTO; break; + case OPT_COPY_UP_SHARED_ON: + config->copy_up_shared = true; + break; + + case OPT_COPY_UP_SHARED_OFF: + config->copy_up_shared = false; + break; + default: pr_err("overlayfs: unrecognized mount option \"%s\" or missing value\n", p); return -EINVAL; @@ -1380,6 +1401,7 @@ static int ovl_fill_super(struct super_block *sb, void *data, int silent) ofs->config.index = ovl_index_def; ofs->config.nfs_export = ovl_nfs_export_def; ofs->config.xino = ovl_xino_def(); + ofs->config.copy_up_shared = ovl_copy_up_shared_def; err = ovl_parse_opt((char *) data, &ofs->config); if (err) goto out_err;
A corner case of a corner case is when - file opened for O_RDONLY - which is then memory mapped SHARED - file opened for O_WRONLY - contents modified - contents read back though the shared mapping Unfortunately it looks very difficult to do anything about the established shared map after the file is copied up. Instead, when a read-only file is mapped shared, copy up the file before actually doing the map. This may result in unnecessary copy-ups (but so may copy-up on open(O_RDWR) for exampe). We can revisit this later if it turns out to be a performance problem in real life. Signed-off-by: Miklos Szeredi <mszeredi@redhat.com> --- fs/overlayfs/Kconfig | 21 +++++++++++++++++++++ fs/overlayfs/file.c | 22 ++++++++++++++++++++++ fs/overlayfs/overlayfs.h | 7 +++++++ fs/overlayfs/ovl_entry.h | 1 + fs/overlayfs/super.c | 22 ++++++++++++++++++++++ 5 files changed, 73 insertions(+)