* [RFC PATCH] ceph: try to prevent exceeding the MDS maximum xattr size
@ 2022-05-20 11:54 Luís Henriques
2022-05-20 14:22 ` Luís Henriques
` (2 more replies)
0 siblings, 3 replies; 7+ messages in thread
From: Luís Henriques @ 2022-05-20 11:54 UTC (permalink / raw)
To: Jeff Layton, Xiubo Li, Ilya Dryomov
Cc: ceph-devel, linux-kernel, Luís Henriques
The MDS tries to enforce a limit on the total key/values in extended
attributes. However, this limit is enforced only if doing a synchronous
operation (MDS_OP_SETXATTR) -- if we're buffering the xattrs, the MDS
doesn't have a chance to enforce these limits.
This patch forces the usage of the synchronous operation if xattrs size hits
the maximum size that is set on the MDS by default (64k).
While there, fix a dout() that would trigger a printk warning:
[ 98.718078] ------------[ cut here ]------------
[ 98.719012] precision 65536 too large
[ 98.719039] WARNING: CPU: 1 PID: 3755 at lib/vsprintf.c:2703 vsnprintf+0x5e3/0x600
...
URL: https://tracker.ceph.com/issues/55725
Signed-off-by: Luís Henriques <lhenriques@suse.de>
---
fs/ceph/xattr.c | 17 +++++++++++++----
1 file changed, 13 insertions(+), 4 deletions(-)
diff --git a/fs/ceph/xattr.c b/fs/ceph/xattr.c
index afec84088471..09751a5f028c 100644
--- a/fs/ceph/xattr.c
+++ b/fs/ceph/xattr.c
@@ -15,6 +15,12 @@
#define XATTR_CEPH_PREFIX "ceph."
#define XATTR_CEPH_PREFIX_LEN (sizeof (XATTR_CEPH_PREFIX) - 1)
+/*
+ * Maximum size of xattrs the MDS can handle per inode by default. This
+ * includes the attribute name and 4+4 bytes for the key/value sizes.
+ */
+#define MDS_MAX_XATTR_PAIRS_SIZE (1<<16) /* 64K */
+
static int __remove_xattr(struct ceph_inode_info *ci,
struct ceph_inode_xattr *xattr);
@@ -1078,7 +1084,7 @@ static int ceph_sync_setxattr(struct inode *inode, const char *name,
flags |= CEPH_XATTR_REMOVE;
}
- dout("setxattr value=%.*s\n", (int)size, value);
+ dout("setxattr value size: ld\n", size);
/* do request */
req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
@@ -1176,8 +1182,13 @@ int __ceph_setxattr(struct inode *inode, const char *name,
spin_lock(&ci->i_ceph_lock);
retry:
issued = __ceph_caps_issued(ci, NULL);
- if (ci->i_xattrs.version == 0 || !(issued & CEPH_CAP_XATTR_EXCL))
+ required_blob_size = __get_required_blob_size(ci, name_len, val_len);
+ if ((ci->i_xattrs.version == 0) || !(issued & CEPH_CAP_XATTR_EXCL) ||
+ (required_blob_size >= MDS_MAX_XATTR_PAIRS_SIZE)) {
+ dout("%s do sync setxattr: version: %llu blob size: %d\n",
+ __func__, ci->i_xattrs.version, required_blob_size);
goto do_sync;
+ }
if (!lock_snap_rwsem && !ci->i_head_snapc) {
lock_snap_rwsem = true;
@@ -1193,8 +1204,6 @@ int __ceph_setxattr(struct inode *inode, const char *name,
ceph_cap_string(issued));
__build_xattrs(inode);
- required_blob_size = __get_required_blob_size(ci, name_len, val_len);
-
if (!ci->i_xattrs.prealloc_blob ||
required_blob_size > ci->i_xattrs.prealloc_blob->alloc_len) {
struct ceph_buffer *blob;
^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [RFC PATCH] ceph: try to prevent exceeding the MDS maximum xattr size
2022-05-20 11:54 [RFC PATCH] ceph: try to prevent exceeding the MDS maximum xattr size Luís Henriques
@ 2022-05-20 14:22 ` Luís Henriques
2022-05-20 16:29 ` kernel test robot
2022-05-23 1:47 ` Xiubo Li
2 siblings, 0 replies; 7+ messages in thread
From: Luís Henriques @ 2022-05-20 14:22 UTC (permalink / raw)
To: Jeff Layton; +Cc: Xiubo Li, Ilya Dryomov, ceph-devel, linux-kernel
Luís Henriques <lhenriques@suse.de> writes:
> The MDS tries to enforce a limit on the total key/values in extended
> attributes. However, this limit is enforced only if doing a synchronous
> operation (MDS_OP_SETXATTR) -- if we're buffering the xattrs, the MDS
> doesn't have a chance to enforce these limits.
>
> This patch forces the usage of the synchronous operation if xattrs size hits
> the maximum size that is set on the MDS by default (64k).
>
> While there, fix a dout() that would trigger a printk warning:
>
> [ 98.718078] ------------[ cut here ]------------
> [ 98.719012] precision 65536 too large
> [ 98.719039] WARNING: CPU: 1 PID: 3755 at lib/vsprintf.c:2703 vsnprintf+0x5e3/0x600
> ...
>
> URL: https://tracker.ceph.com/issues/55725
> Signed-off-by: Luís Henriques <lhenriques@suse.de>
> ---
> fs/ceph/xattr.c | 17 +++++++++++++----
> 1 file changed, 13 insertions(+), 4 deletions(-)
>
> diff --git a/fs/ceph/xattr.c b/fs/ceph/xattr.c
> index afec84088471..09751a5f028c 100644
> --- a/fs/ceph/xattr.c
> +++ b/fs/ceph/xattr.c
> @@ -15,6 +15,12 @@
> #define XATTR_CEPH_PREFIX "ceph."
> #define XATTR_CEPH_PREFIX_LEN (sizeof (XATTR_CEPH_PREFIX) - 1)
>
> +/*
> + * Maximum size of xattrs the MDS can handle per inode by default. This
> + * includes the attribute name and 4+4 bytes for the key/value sizes.
> + */
> +#define MDS_MAX_XATTR_PAIRS_SIZE (1<<16) /* 64K */
> +
> static int __remove_xattr(struct ceph_inode_info *ci,
> struct ceph_inode_xattr *xattr);
>
> @@ -1078,7 +1084,7 @@ static int ceph_sync_setxattr(struct inode *inode, const char *name,
> flags |= CEPH_XATTR_REMOVE;
> }
>
> - dout("setxattr value=%.*s\n", (int)size, value);
> + dout("setxattr value size: ld\n", size);
Oops! Looks like someone ate a '%' char. Oh well, I won't bother sending
out a new version for now as this is an RFC and the MDS side is what
really needs fixing. In fact, the client-side may be something very
different from this RFC.
Cheers,
--
Luís
>
> /* do request */
> req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
> @@ -1176,8 +1182,13 @@ int __ceph_setxattr(struct inode *inode, const char *name,
> spin_lock(&ci->i_ceph_lock);
> retry:
> issued = __ceph_caps_issued(ci, NULL);
> - if (ci->i_xattrs.version == 0 || !(issued & CEPH_CAP_XATTR_EXCL))
> + required_blob_size = __get_required_blob_size(ci, name_len, val_len);
> + if ((ci->i_xattrs.version == 0) || !(issued & CEPH_CAP_XATTR_EXCL) ||
> + (required_blob_size >= MDS_MAX_XATTR_PAIRS_SIZE)) {
> + dout("%s do sync setxattr: version: %llu blob size: %d\n",
> + __func__, ci->i_xattrs.version, required_blob_size);
> goto do_sync;
> + }
>
> if (!lock_snap_rwsem && !ci->i_head_snapc) {
> lock_snap_rwsem = true;
> @@ -1193,8 +1204,6 @@ int __ceph_setxattr(struct inode *inode, const char *name,
> ceph_cap_string(issued));
> __build_xattrs(inode);
>
> - required_blob_size = __get_required_blob_size(ci, name_len, val_len);
> -
> if (!ci->i_xattrs.prealloc_blob ||
> required_blob_size > ci->i_xattrs.prealloc_blob->alloc_len) {
> struct ceph_buffer *blob;
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [RFC PATCH] ceph: try to prevent exceeding the MDS maximum xattr size
2022-05-20 11:54 [RFC PATCH] ceph: try to prevent exceeding the MDS maximum xattr size Luís Henriques
2022-05-20 14:22 ` Luís Henriques
@ 2022-05-20 16:29 ` kernel test robot
2022-05-23 1:47 ` Xiubo Li
2 siblings, 0 replies; 7+ messages in thread
From: kernel test robot @ 2022-05-20 16:29 UTC (permalink / raw)
To: Luís Henriques; +Cc: llvm, kbuild-all
Hi "Luís,
[FYI, it's a private test report for your RFC patch.]
[auto build test WARNING on ceph-client/for-linus]
[also build test WARNING on v5.18-rc7 next-20220520]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]
url: https://github.com/intel-lab-lkp/linux/commits/Lu-s-Henriques/ceph-try-to-prevent-exceeding-the-MDS-maximum-xattr-size/20220520-195559
base: https://github.com/ceph/ceph-client.git for-linus
config: i386-randconfig-a006 (https://download.01.org/0day-ci/archive/20220521/202205210021.IgsLVawx-lkp@intel.com/config)
compiler: clang version 15.0.0 (https://github.com/llvm/llvm-project e00cbbec06c08dc616a0d52a20f678b8fbd4e304)
reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
chmod +x ~/bin/make.cross
# https://github.com/intel-lab-lkp/linux/commit/3c9e88c091f0e9b17856d7de9b8c705e827ee716
git remote add linux-review https://github.com/intel-lab-lkp/linux
git fetch --no-tags linux-review Lu-s-Henriques/ceph-try-to-prevent-exceeding-the-MDS-maximum-xattr-size/20220520-195559
git checkout 3c9e88c091f0e9b17856d7de9b8c705e827ee716
# save the config file
mkdir build_dir && cp config build_dir/.config
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross W=1 O=build_dir ARCH=i386 SHELL=/bin/bash fs/ceph/
If you fix the issue, kindly add following tag where applicable
Reported-by: kernel test robot <lkp@intel.com>
All warnings (new ones prefixed by >>):
>> fs/ceph/xattr.c:1087:36: warning: data argument not used by format string [-Wformat-extra-args]
dout("setxattr value size: ld\n", size);
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~
include/linux/ceph/ceph_debug.h:35:45: note: expanded from macro 'dout'
# define dout(fmt, ...) pr_debug(" " fmt, ##__VA_ARGS__)
~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~
include/linux/printk.h:576:38: note: expanded from macro 'pr_debug'
no_printk(KERN_DEBUG pr_fmt(fmt), ##__VA_ARGS__)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~
include/linux/printk.h:132:17: note: expanded from macro 'no_printk'
printk(fmt, ##__VA_ARGS__); \
~~~~~~~~~~~~~~^~~~~~~~~~~~
include/linux/printk.h:446:60: note: expanded from macro 'printk'
#define printk(fmt, ...) printk_index_wrap(_printk, fmt, ##__VA_ARGS__)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~
include/linux/printk.h:418:19: note: expanded from macro 'printk_index_wrap'
_p_func(_fmt, ##__VA_ARGS__); \
~~~~ ^
1 warning generated.
vim +1087 fs/ceph/xattr.c
1058
1059 static int ceph_sync_setxattr(struct inode *inode, const char *name,
1060 const char *value, size_t size, int flags)
1061 {
1062 struct ceph_fs_client *fsc = ceph_sb_to_client(inode->i_sb);
1063 struct ceph_inode_info *ci = ceph_inode(inode);
1064 struct ceph_mds_request *req;
1065 struct ceph_mds_client *mdsc = fsc->mdsc;
1066 struct ceph_osd_client *osdc = &fsc->client->osdc;
1067 struct ceph_pagelist *pagelist = NULL;
1068 int op = CEPH_MDS_OP_SETXATTR;
1069 int err;
1070
1071 if (size > 0) {
1072 /* copy value into pagelist */
1073 pagelist = ceph_pagelist_alloc(GFP_NOFS);
1074 if (!pagelist)
1075 return -ENOMEM;
1076
1077 err = ceph_pagelist_append(pagelist, value, size);
1078 if (err)
1079 goto out;
1080 } else if (!value) {
1081 if (flags & CEPH_XATTR_REPLACE)
1082 op = CEPH_MDS_OP_RMXATTR;
1083 else
1084 flags |= CEPH_XATTR_REMOVE;
1085 }
1086
> 1087 dout("setxattr value size: ld\n", size);
1088
1089 /* do request */
1090 req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
1091 if (IS_ERR(req)) {
1092 err = PTR_ERR(req);
1093 goto out;
1094 }
1095
1096 req->r_path2 = kstrdup(name, GFP_NOFS);
1097 if (!req->r_path2) {
1098 ceph_mdsc_put_request(req);
1099 err = -ENOMEM;
1100 goto out;
1101 }
1102
1103 if (op == CEPH_MDS_OP_SETXATTR) {
1104 req->r_args.setxattr.flags = cpu_to_le32(flags);
1105 req->r_args.setxattr.osdmap_epoch =
1106 cpu_to_le32(osdc->osdmap->epoch);
1107 req->r_pagelist = pagelist;
1108 pagelist = NULL;
1109 }
1110
1111 req->r_inode = inode;
1112 ihold(inode);
1113 req->r_num_caps = 1;
1114 req->r_inode_drop = CEPH_CAP_XATTR_SHARED;
1115
1116 dout("xattr.ver (before): %lld\n", ci->i_xattrs.version);
1117 err = ceph_mdsc_do_request(mdsc, NULL, req);
1118 ceph_mdsc_put_request(req);
1119 dout("xattr.ver (after): %lld\n", ci->i_xattrs.version);
1120
1121 out:
1122 if (pagelist)
1123 ceph_pagelist_release(pagelist);
1124 return err;
1125 }
1126
--
0-DAY CI Kernel Test Service
https://01.org/lkp
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [RFC PATCH] ceph: try to prevent exceeding the MDS maximum xattr size
2022-05-20 11:54 [RFC PATCH] ceph: try to prevent exceeding the MDS maximum xattr size Luís Henriques
2022-05-20 14:22 ` Luís Henriques
2022-05-20 16:29 ` kernel test robot
@ 2022-05-23 1:47 ` Xiubo Li
2022-05-23 9:47 ` Luís Henriques
2022-05-23 10:43 ` Jeff Layton
2 siblings, 2 replies; 7+ messages in thread
From: Xiubo Li @ 2022-05-23 1:47 UTC (permalink / raw)
To: Luís Henriques, Jeff Layton, Ilya Dryomov; +Cc: ceph-devel, linux-kernel
On 5/20/22 7:54 PM, Luís Henriques wrote:
> The MDS tries to enforce a limit on the total key/values in extended
> attributes. However, this limit is enforced only if doing a synchronous
> operation (MDS_OP_SETXATTR) -- if we're buffering the xattrs, the MDS
> doesn't have a chance to enforce these limits.
>
> This patch forces the usage of the synchronous operation if xattrs size hits
> the maximum size that is set on the MDS by default (64k).
>
> While there, fix a dout() that would trigger a printk warning:
>
> [ 98.718078] ------------[ cut here ]------------
> [ 98.719012] precision 65536 too large
> [ 98.719039] WARNING: CPU: 1 PID: 3755 at lib/vsprintf.c:2703 vsnprintf+0x5e3/0x600
> ...
>
> URL: https://tracker.ceph.com/issues/55725
> Signed-off-by: Luís Henriques <lhenriques@suse.de>
> ---
> fs/ceph/xattr.c | 17 +++++++++++++----
> 1 file changed, 13 insertions(+), 4 deletions(-)
>
> diff --git a/fs/ceph/xattr.c b/fs/ceph/xattr.c
> index afec84088471..09751a5f028c 100644
> --- a/fs/ceph/xattr.c
> +++ b/fs/ceph/xattr.c
> @@ -15,6 +15,12 @@
> #define XATTR_CEPH_PREFIX "ceph."
> #define XATTR_CEPH_PREFIX_LEN (sizeof (XATTR_CEPH_PREFIX) - 1)
>
> +/*
> + * Maximum size of xattrs the MDS can handle per inode by default. This
> + * includes the attribute name and 4+4 bytes for the key/value sizes.
> + */
> +#define MDS_MAX_XATTR_PAIRS_SIZE (1<<16) /* 64K */
The max size is changeable in MDS side. Possibly we should do something
as mentioned in your ceph PR [1].
@Jeff, any better idea ?
[1]
https://github.com/ceph/ceph/pull/46357/commits/741f8ba36f14774834c0d5618519425ccf1ccc85#r878966753
Thanks.
-- Xiubo
> +
> static int __remove_xattr(struct ceph_inode_info *ci,
> struct ceph_inode_xattr *xattr);
>
> @@ -1078,7 +1084,7 @@ static int ceph_sync_setxattr(struct inode *inode, const char *name,
> flags |= CEPH_XATTR_REMOVE;
> }
>
> - dout("setxattr value=%.*s\n", (int)size, value);
> + dout("setxattr value size: ld\n", size);
>
> /* do request */
> req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
> @@ -1176,8 +1182,13 @@ int __ceph_setxattr(struct inode *inode, const char *name,
> spin_lock(&ci->i_ceph_lock);
> retry:
> issued = __ceph_caps_issued(ci, NULL);
> - if (ci->i_xattrs.version == 0 || !(issued & CEPH_CAP_XATTR_EXCL))
> + required_blob_size = __get_required_blob_size(ci, name_len, val_len);
> + if ((ci->i_xattrs.version == 0) || !(issued & CEPH_CAP_XATTR_EXCL) ||
> + (required_blob_size >= MDS_MAX_XATTR_PAIRS_SIZE)) {
> + dout("%s do sync setxattr: version: %llu blob size: %d\n",
> + __func__, ci->i_xattrs.version, required_blob_size);
> goto do_sync;
> + }
>
> if (!lock_snap_rwsem && !ci->i_head_snapc) {
> lock_snap_rwsem = true;
> @@ -1193,8 +1204,6 @@ int __ceph_setxattr(struct inode *inode, const char *name,
> ceph_cap_string(issued));
> __build_xattrs(inode);
>
> - required_blob_size = __get_required_blob_size(ci, name_len, val_len);
> -
> if (!ci->i_xattrs.prealloc_blob ||
> required_blob_size > ci->i_xattrs.prealloc_blob->alloc_len) {
> struct ceph_buffer *blob;
>
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [RFC PATCH] ceph: try to prevent exceeding the MDS maximum xattr size
2022-05-23 1:47 ` Xiubo Li
@ 2022-05-23 9:47 ` Luís Henriques
2022-05-23 10:43 ` Jeff Layton
1 sibling, 0 replies; 7+ messages in thread
From: Luís Henriques @ 2022-05-23 9:47 UTC (permalink / raw)
To: Xiubo Li; +Cc: Jeff Layton, Ilya Dryomov, ceph-devel, linux-kernel
Xiubo Li <xiubli@redhat.com> writes:
> On 5/20/22 7:54 PM, Luís Henriques wrote:
>> The MDS tries to enforce a limit on the total key/values in extended
>> attributes. However, this limit is enforced only if doing a synchronous
>> operation (MDS_OP_SETXATTR) -- if we're buffering the xattrs, the MDS
>> doesn't have a chance to enforce these limits.
>>
>> This patch forces the usage of the synchronous operation if xattrs size hits
>> the maximum size that is set on the MDS by default (64k).
>>
>> While there, fix a dout() that would trigger a printk warning:
>>
>> [ 98.718078] ------------[ cut here ]------------
>> [ 98.719012] precision 65536 too large
>> [ 98.719039] WARNING: CPU: 1 PID: 3755 at lib/vsprintf.c:2703 vsnprintf+0x5e3/0x600
>> ...
>>
>> URL: https://tracker.ceph.com/issues/55725
>> Signed-off-by: Luís Henriques <lhenriques@suse.de>
>> ---
>> fs/ceph/xattr.c | 17 +++++++++++++----
>> 1 file changed, 13 insertions(+), 4 deletions(-)
>>
>> diff --git a/fs/ceph/xattr.c b/fs/ceph/xattr.c
>> index afec84088471..09751a5f028c 100644
>> --- a/fs/ceph/xattr.c
>> +++ b/fs/ceph/xattr.c
>> @@ -15,6 +15,12 @@
>> #define XATTR_CEPH_PREFIX "ceph."
>> #define XATTR_CEPH_PREFIX_LEN (sizeof (XATTR_CEPH_PREFIX) - 1)
>> +/*
>> + * Maximum size of xattrs the MDS can handle per inode by default. This
>> + * includes the attribute name and 4+4 bytes for the key/value sizes.
>> + */
>> +#define MDS_MAX_XATTR_PAIRS_SIZE (1<<16) /* 64K */
>
> The max size is changeable in MDS side. Possibly we should do something as
> mentioned in your ceph PR [1].
>
Right, as I mentioned in that PR, having clients that are aware of that
limit will allow them to behave correctly. But the MDS still needs to
handle the case where these limits are exceeded (for ex. kernel clients
that don't know about this limit). So, we still need to do something
similar to what I've proposed in there.
Cheers
--
Luís
> @Jeff, any better idea ?
>
>
> [1]
> https://github.com/ceph/ceph/pull/46357/commits/741f8ba36f14774834c0d5618519425ccf1ccc85#r878966753
>
> Thanks.
>
> -- Xiubo
>
>
>> +
>> static int __remove_xattr(struct ceph_inode_info *ci,
>> struct ceph_inode_xattr *xattr);
>> @@ -1078,7 +1084,7 @@ static int ceph_sync_setxattr(struct inode *inode,
>> const char *name,
>> flags |= CEPH_XATTR_REMOVE;
>> }
>> - dout("setxattr value=%.*s\n", (int)size, value);
>> + dout("setxattr value size: ld\n", size);
>> /* do request */
>> req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
>> @@ -1176,8 +1182,13 @@ int __ceph_setxattr(struct inode *inode, const char *name,
>> spin_lock(&ci->i_ceph_lock);
>> retry:
>> issued = __ceph_caps_issued(ci, NULL);
>> - if (ci->i_xattrs.version == 0 || !(issued & CEPH_CAP_XATTR_EXCL))
>> + required_blob_size = __get_required_blob_size(ci, name_len, val_len);
>> + if ((ci->i_xattrs.version == 0) || !(issued & CEPH_CAP_XATTR_EXCL) ||
>> + (required_blob_size >= MDS_MAX_XATTR_PAIRS_SIZE)) {
>> + dout("%s do sync setxattr: version: %llu blob size: %d\n",
>> + __func__, ci->i_xattrs.version, required_blob_size);
>> goto do_sync;
>> + }
>> if (!lock_snap_rwsem && !ci->i_head_snapc) {
>> lock_snap_rwsem = true;
>> @@ -1193,8 +1204,6 @@ int __ceph_setxattr(struct inode *inode, const char *name,
>> ceph_cap_string(issued));
>> __build_xattrs(inode);
>> - required_blob_size = __get_required_blob_size(ci, name_len, val_len);
>> -
>> if (!ci->i_xattrs.prealloc_blob ||
>> required_blob_size > ci->i_xattrs.prealloc_blob->alloc_len) {
>> struct ceph_buffer *blob;
>>
>
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [RFC PATCH] ceph: try to prevent exceeding the MDS maximum xattr size
2022-05-23 1:47 ` Xiubo Li
2022-05-23 9:47 ` Luís Henriques
@ 2022-05-23 10:43 ` Jeff Layton
2022-05-23 12:11 ` Xiubo Li
1 sibling, 1 reply; 7+ messages in thread
From: Jeff Layton @ 2022-05-23 10:43 UTC (permalink / raw)
To: Xiubo Li, Luís Henriques, Ilya Dryomov; +Cc: ceph-devel, linux-kernel
On Mon, 2022-05-23 at 09:47 +0800, Xiubo Li wrote:
> On 5/20/22 7:54 PM, Luís Henriques wrote:
> > The MDS tries to enforce a limit on the total key/values in extended
> > attributes. However, this limit is enforced only if doing a synchronous
> > operation (MDS_OP_SETXATTR) -- if we're buffering the xattrs, the MDS
> > doesn't have a chance to enforce these limits.
> >
> > This patch forces the usage of the synchronous operation if xattrs size hits
> > the maximum size that is set on the MDS by default (64k).
> >
> > While there, fix a dout() that would trigger a printk warning:
> >
> > [ 98.718078] ------------[ cut here ]------------
> > [ 98.719012] precision 65536 too large
> > [ 98.719039] WARNING: CPU: 1 PID: 3755 at lib/vsprintf.c:2703 vsnprintf+0x5e3/0x600
> > ...
> >
> > URL: https://tracker.ceph.com/issues/55725
> > Signed-off-by: Luís Henriques <lhenriques@suse.de>
> > ---
> > fs/ceph/xattr.c | 17 +++++++++++++----
> > 1 file changed, 13 insertions(+), 4 deletions(-)
> >
> > diff --git a/fs/ceph/xattr.c b/fs/ceph/xattr.c
> > index afec84088471..09751a5f028c 100644
> > --- a/fs/ceph/xattr.c
> > +++ b/fs/ceph/xattr.c
> > @@ -15,6 +15,12 @@
> > #define XATTR_CEPH_PREFIX "ceph."
> > #define XATTR_CEPH_PREFIX_LEN (sizeof (XATTR_CEPH_PREFIX) - 1)
> >
> > +/*
> > + * Maximum size of xattrs the MDS can handle per inode by default. This
> > + * includes the attribute name and 4+4 bytes for the key/value sizes.
> > + */
> > +#define MDS_MAX_XATTR_PAIRS_SIZE (1<<16) /* 64K */
>
> The max size is changeable in MDS side. Possibly we should do something
> as mentioned in your ceph PR [1].
>
> @Jeff, any better idea ?
>
>
> [1]
> https://github.com/ceph/ceph/pull/46357/commits/741f8ba36f14774834c0d5618519425ccf1ccc85#r878966753
>
> Thanks.
>
> -- Xiubo
>
>
Not really.
The idea in the github comment makes sense. Basically, make it so that
the setting isn't changeable at runtime and then have the client query
for the limit at appropriate times.
You can probably still defeat that by changing it and rebooting the MDS,
but I don't see that you can do much else.
> > +
> > static int __remove_xattr(struct ceph_inode_info *ci,
> > struct ceph_inode_xattr *xattr);
> >
> > @@ -1078,7 +1084,7 @@ static int ceph_sync_setxattr(struct inode *inode, const char *name,
> > flags |= CEPH_XATTR_REMOVE;
> > }
> >
> > - dout("setxattr value=%.*s\n", (int)size, value);
> > + dout("setxattr value size: ld\n", size);
> >
> > /* do request */
> > req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
> > @@ -1176,8 +1182,13 @@ int __ceph_setxattr(struct inode *inode, const char *name,
> > spin_lock(&ci->i_ceph_lock);
> > retry:
> > issued = __ceph_caps_issued(ci, NULL);
> > - if (ci->i_xattrs.version == 0 || !(issued & CEPH_CAP_XATTR_EXCL))
> > + required_blob_size = __get_required_blob_size(ci, name_len, val_len);
> > + if ((ci->i_xattrs.version == 0) || !(issued & CEPH_CAP_XATTR_EXCL) ||
> > + (required_blob_size >= MDS_MAX_XATTR_PAIRS_SIZE)) {
> > + dout("%s do sync setxattr: version: %llu blob size: %d\n",
> > + __func__, ci->i_xattrs.version, required_blob_size);
> > goto do_sync;
> > + }
> >
> > if (!lock_snap_rwsem && !ci->i_head_snapc) {
> > lock_snap_rwsem = true;
> > @@ -1193,8 +1204,6 @@ int __ceph_setxattr(struct inode *inode, const char *name,
> > ceph_cap_string(issued));
> > __build_xattrs(inode);
> >
> > - required_blob_size = __get_required_blob_size(ci, name_len, val_len);
> > -
> > if (!ci->i_xattrs.prealloc_blob ||
> > required_blob_size > ci->i_xattrs.prealloc_blob->alloc_len) {
> > struct ceph_buffer *blob;
> >
>
--
Jeff Layton <jlayton@kernel.org>
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [RFC PATCH] ceph: try to prevent exceeding the MDS maximum xattr size
2022-05-23 10:43 ` Jeff Layton
@ 2022-05-23 12:11 ` Xiubo Li
0 siblings, 0 replies; 7+ messages in thread
From: Xiubo Li @ 2022-05-23 12:11 UTC (permalink / raw)
To: Jeff Layton, Luís Henriques, Ilya Dryomov; +Cc: ceph-devel, linux-kernel
On 5/23/22 6:43 PM, Jeff Layton wrote:
> On Mon, 2022-05-23 at 09:47 +0800, Xiubo Li wrote:
>> On 5/20/22 7:54 PM, Luís Henriques wrote:
>>> The MDS tries to enforce a limit on the total key/values in extended
>>> attributes. However, this limit is enforced only if doing a synchronous
>>> operation (MDS_OP_SETXATTR) -- if we're buffering the xattrs, the MDS
>>> doesn't have a chance to enforce these limits.
>>>
>>> This patch forces the usage of the synchronous operation if xattrs size hits
>>> the maximum size that is set on the MDS by default (64k).
>>>
>>> While there, fix a dout() that would trigger a printk warning:
>>>
>>> [ 98.718078] ------------[ cut here ]------------
>>> [ 98.719012] precision 65536 too large
>>> [ 98.719039] WARNING: CPU: 1 PID: 3755 at lib/vsprintf.c:2703 vsnprintf+0x5e3/0x600
>>> ...
>>>
>>> URL: https://tracker.ceph.com/issues/55725
>>> Signed-off-by: Luís Henriques <lhenriques@suse.de>
>>> ---
>>> fs/ceph/xattr.c | 17 +++++++++++++----
>>> 1 file changed, 13 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/fs/ceph/xattr.c b/fs/ceph/xattr.c
>>> index afec84088471..09751a5f028c 100644
>>> --- a/fs/ceph/xattr.c
>>> +++ b/fs/ceph/xattr.c
>>> @@ -15,6 +15,12 @@
>>> #define XATTR_CEPH_PREFIX "ceph."
>>> #define XATTR_CEPH_PREFIX_LEN (sizeof (XATTR_CEPH_PREFIX) - 1)
>>>
>>> +/*
>>> + * Maximum size of xattrs the MDS can handle per inode by default. This
>>> + * includes the attribute name and 4+4 bytes for the key/value sizes.
>>> + */
>>> +#define MDS_MAX_XATTR_PAIRS_SIZE (1<<16) /* 64K */
>> The max size is changeable in MDS side. Possibly we should do something
>> as mentioned in your ceph PR [1].
>>
>> @Jeff, any better idea ?
>>
>>
>> [1]
>> https://github.com/ceph/ceph/pull/46357/commits/741f8ba36f14774834c0d5618519425ccf1ccc85#r878966753
>>
>> Thanks.
>>
>> -- Xiubo
>>
>>
> Not really.
>
> The idea in the github comment makes sense. Basically, make it so that
> the setting isn't changeable at runtime and then have the client query
> for the limit at appropriate times.
>
> You can probably still defeat that by changing it and rebooting the MDS,
> but I don't see that you can do much else.
Yeah, sounds good.
>
>>> +
>>> static int __remove_xattr(struct ceph_inode_info *ci,
>>> struct ceph_inode_xattr *xattr);
>>>
>>> @@ -1078,7 +1084,7 @@ static int ceph_sync_setxattr(struct inode *inode, const char *name,
>>> flags |= CEPH_XATTR_REMOVE;
>>> }
>>>
>>> - dout("setxattr value=%.*s\n", (int)size, value);
>>> + dout("setxattr value size: ld\n", size);
>>>
>>> /* do request */
>>> req = ceph_mdsc_create_request(mdsc, op, USE_AUTH_MDS);
>>> @@ -1176,8 +1182,13 @@ int __ceph_setxattr(struct inode *inode, const char *name,
>>> spin_lock(&ci->i_ceph_lock);
>>> retry:
>>> issued = __ceph_caps_issued(ci, NULL);
>>> - if (ci->i_xattrs.version == 0 || !(issued & CEPH_CAP_XATTR_EXCL))
>>> + required_blob_size = __get_required_blob_size(ci, name_len, val_len);
>>> + if ((ci->i_xattrs.version == 0) || !(issued & CEPH_CAP_XATTR_EXCL) ||
>>> + (required_blob_size >= MDS_MAX_XATTR_PAIRS_SIZE)) {
>>> + dout("%s do sync setxattr: version: %llu blob size: %d\n",
>>> + __func__, ci->i_xattrs.version, required_blob_size);
>>> goto do_sync;
>>> + }
>>>
>>> if (!lock_snap_rwsem && !ci->i_head_snapc) {
>>> lock_snap_rwsem = true;
>>> @@ -1193,8 +1204,6 @@ int __ceph_setxattr(struct inode *inode, const char *name,
>>> ceph_cap_string(issued));
>>> __build_xattrs(inode);
>>>
>>> - required_blob_size = __get_required_blob_size(ci, name_len, val_len);
>>> -
>>> if (!ci->i_xattrs.prealloc_blob ||
>>> required_blob_size > ci->i_xattrs.prealloc_blob->alloc_len) {
>>> struct ceph_buffer *blob;
>>>
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2022-05-23 12:11 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-05-20 11:54 [RFC PATCH] ceph: try to prevent exceeding the MDS maximum xattr size Luís Henriques
2022-05-20 14:22 ` Luís Henriques
2022-05-20 16:29 ` kernel test robot
2022-05-23 1:47 ` Xiubo Li
2022-05-23 9:47 ` Luís Henriques
2022-05-23 10:43 ` Jeff Layton
2022-05-23 12:11 ` Xiubo Li
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.