| Message ID | 20240612011156.2891254-3-lihongbo22@huawei.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | Introduce tracepoint for hugetlbfs | expand |
Just a friendly ping to the patch :) https://lore.kernel.org/all/20240612011156.2891254-1-lihongbo22@huawei.com/ Thanks, Hongbo On 2024/6/12 9:11, Hongbo Li wrote: > Here we use the hugetlbfs tracepoint to track the call stack. And > the output in trace is as follows: > > ``` > touch-5307 [004] ..... 1402.167607: hugetlbfs_alloc_inode: dev = (0,50), ino = 21380, dir = 16921, mode = 0100644 > touch-5307 [004] ..... 1402.167638: hugetlbfs_setattr: dev = (0,50), ino = 21380, name = testfile1, ia_valid = 131184, ia_mode = 0132434, ia_uid = 2863018275, ia_gid = 4294967295, old_size = 0, ia_size = 4064 > truncate-5328 [003] ..... 1436.031054: hugetlbfs_setattr: dev = (0,50), ino = 21380, name = testfile1, ia_valid = 8296, ia_mode = 0177777, ia_uid = 2862574544, ia_gid = 4294967295, old_size = 0, ia_size = 2097152 > rm-5338 [004] ..... 1484.426247: hugetlbfs_evict_inode: dev = (0,50), ino = 21380, i_mode = 0100644, i_size = 2097152, i_nlink = 0, seals = 1, i_blocks = 0 > <idle>-0 [004] ..s1. 1484.446668: hugetlbfs_free_inode: dev = (0,50), ino = 21380, i_mode = 0100644, i_size = 2097152, i_nlink = 0, seals = 1, i_blocks = 0 > ``` > > Signed-off-by: Hongbo Li <lihongbo22@huawei.com> > --- > fs/hugetlbfs/inode.c | 21 +++++++++++++++++++-- > 1 file changed, 19 insertions(+), 2 deletions(-) > > diff --git a/fs/hugetlbfs/inode.c b/fs/hugetlbfs/inode.c > index 412f295acebe..f3399c6a02ca 100644 > --- a/fs/hugetlbfs/inode.c > +++ b/fs/hugetlbfs/inode.c > @@ -39,6 +39,9 @@ > #include <linux/uaccess.h> > #include <linux/sched/mm.h> > > +#define CREATE_TRACE_POINTS > +#include <trace/events/hugetlbfs.h> > + > static const struct address_space_operations hugetlbfs_aops; > static const struct file_operations hugetlbfs_file_operations; > static const struct inode_operations hugetlbfs_dir_inode_operations; > @@ -686,6 +689,7 @@ static void hugetlbfs_evict_inode(struct inode *inode) > { > struct resv_map *resv_map; > > + trace_hugetlbfs_evict_inode(inode); > remove_inode_hugepages(inode, 0, LLONG_MAX); > > /* > @@ -813,8 +817,10 @@ static long hugetlbfs_fallocate(struct file *file, int mode, loff_t offset, > if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE)) > return -EOPNOTSUPP; > > - if (mode & FALLOC_FL_PUNCH_HOLE) > - return hugetlbfs_punch_hole(inode, offset, len); > + if (mode & FALLOC_FL_PUNCH_HOLE) { > + error = hugetlbfs_punch_hole(inode, offset, len); > + goto out_nolock; > + } > > /* > * Default preallocate case. > @@ -918,6 +924,9 @@ static long hugetlbfs_fallocate(struct file *file, int mode, loff_t offset, > inode_set_ctime_current(inode); > out: > inode_unlock(inode); > + > +out_nolock: > + trace_hugetlbfs_fallocate(inode, mode, offset, len, error); > return error; > } > > @@ -934,6 +943,12 @@ static int hugetlbfs_setattr(struct mnt_idmap *idmap, > if (error) > return error; > > + trace_hugetlbfs_setattr(inode, dentry->d_name.len, dentry->d_name.name, > + attr->ia_valid, attr->ia_mode, > + from_kuid(&init_user_ns, attr->ia_uid), > + from_kgid(&init_user_ns, attr->ia_gid), > + inode->i_size, attr->ia_size); > + > if (ia_valid & ATTR_SIZE) { > loff_t oldsize = inode->i_size; > loff_t newsize = attr->ia_size; > @@ -1032,6 +1047,7 @@ static struct inode *hugetlbfs_get_inode(struct super_block *sb, > break; > } > lockdep_annotate_inode_mutex_key(inode); > + trace_hugetlbfs_alloc_inode(inode, dir, mode); > } else { > if (resv_map) > kref_put(&resv_map->refs, resv_map_release); > @@ -1274,6 +1290,7 @@ static struct inode *hugetlbfs_alloc_inode(struct super_block *sb) > > static void hugetlbfs_free_inode(struct inode *inode) > { > + trace_hugetlbfs_free_inode(inode); > kmem_cache_free(hugetlbfs_inode_cachep, HUGETLBFS_I(inode)); > } >
On Wed, 12 Jun 2024 09:11:56 +0800 Hongbo Li <lihongbo22@huawei.com> wrote: > @@ -934,6 +943,12 @@ static int hugetlbfs_setattr(struct mnt_idmap *idmap, > if (error) > return error; > > + trace_hugetlbfs_setattr(inode, dentry->d_name.len, dentry->d_name.name, > + attr->ia_valid, attr->ia_mode, > + from_kuid(&init_user_ns, attr->ia_uid), > + from_kgid(&init_user_ns, attr->ia_gid), > + inode->i_size, attr->ia_size); > + That's a lot of parameters to pass to a tracepoint. Why not just pass the dentry and attr and do the above in the TP_fast_assign() logic? That would put less pressure on the icache for the code part. -- Steve
On 2024/7/2 7:49, Steven Rostedt wrote: > On Wed, 12 Jun 2024 09:11:56 +0800 > Hongbo Li <lihongbo22@huawei.com> wrote: > >> @@ -934,6 +943,12 @@ static int hugetlbfs_setattr(struct mnt_idmap *idmap, >> if (error) >> return error; >> >> + trace_hugetlbfs_setattr(inode, dentry->d_name.len, dentry->d_name.name, >> + attr->ia_valid, attr->ia_mode, >> + from_kuid(&init_user_ns, attr->ia_uid), >> + from_kgid(&init_user_ns, attr->ia_gid), >> + inode->i_size, attr->ia_size); >> + > > That's a lot of parameters to pass to a tracepoint. Why not just pass the > dentry and attr and do the above in the TP_fast_assign() logic? That would > put less pressure on the icache for the code part. Thanks for reviewing! Some logic such as kuid_t --> uid_t might be reasonable obtained in filesystem layer. Passing the dentry and attr will let trace know the meaning of structure, perhaps tracepoint should not be aware of the members of these structures as much as possible. Thanks, Hongbo > > -- Steve >
On 2024-07-02 07:55, Hongbo Li wrote: > > > On 2024/7/2 7:49, Steven Rostedt wrote: >> On Wed, 12 Jun 2024 09:11:56 +0800 >> Hongbo Li <lihongbo22@huawei.com> wrote: >> >>> @@ -934,6 +943,12 @@ static int hugetlbfs_setattr(struct mnt_idmap >>> *idmap, >>> if (error) >>> return error; >>> + trace_hugetlbfs_setattr(inode, dentry->d_name.len, >>> dentry->d_name.name, >>> + attr->ia_valid, attr->ia_mode, >>> + from_kuid(&init_user_ns, attr->ia_uid), >>> + from_kgid(&init_user_ns, attr->ia_gid), >>> + inode->i_size, attr->ia_size); >>> + >> >> That's a lot of parameters to pass to a tracepoint. Why not just pass the >> dentry and attr and do the above in the TP_fast_assign() logic? That >> would >> put less pressure on the icache for the code part. > > Thanks for reviewing! > > Some logic such as kuid_t --> uid_t might be reasonable obtained in > filesystem layer. Passing the dentry and attr will let trace know the > meaning of structure, perhaps tracepoint should not be aware of the > members of these structures as much as possible. As maintainer of the LTTng out-of-tree kernel tracer, I appreciate the effort to decouple instrumentation from the subsystem instrumentation, but as long as the structure sits in public headers and the global variables used within the TP_fast_assign() logic (e.g. init_user_ns) are export-gpl, this is enough to make it easy for tracer integration and it keeps the tracepoint caller code footprint to a minimum. The TRACE_EVENT definitions are specific to the subsystem anyway, so I don't think it matters that the TRACE_EVENT() need to access the dentry and attr structures. So I agree with Steven's suggestion. However, just as a precision, I suspect it will have mainly an impact on code size, but not necessarily on icache footprint, because it will shrink the code size within the tracepoint unlikely branch (cold instructions). Thanks, Mathieu > > Thanks, > Hongbo > >> >> -- Steve >>
On 2024/7/2 21:30, Mathieu Desnoyers wrote: > On 2024-07-02 07:55, Hongbo Li wrote: >> >> >> On 2024/7/2 7:49, Steven Rostedt wrote: >>> On Wed, 12 Jun 2024 09:11:56 +0800 >>> Hongbo Li <lihongbo22@huawei.com> wrote: >>> >>>> @@ -934,6 +943,12 @@ static int hugetlbfs_setattr(struct mnt_idmap >>>> *idmap, >>>> if (error) >>>> return error; >>>> + trace_hugetlbfs_setattr(inode, dentry->d_name.len, >>>> dentry->d_name.name, >>>> + attr->ia_valid, attr->ia_mode, >>>> + from_kuid(&init_user_ns, attr->ia_uid), >>>> + from_kgid(&init_user_ns, attr->ia_gid), >>>> + inode->i_size, attr->ia_size); >>>> + >>> >>> That's a lot of parameters to pass to a tracepoint. Why not just pass >>> the >>> dentry and attr and do the above in the TP_fast_assign() logic? That >>> would >>> put less pressure on the icache for the code part. >> >> Thanks for reviewing! >> >> Some logic such as kuid_t --> uid_t might be reasonable obtained in >> filesystem layer. Passing the dentry and attr will let trace know the >> meaning of structure, perhaps tracepoint should not be aware of the >> members of these structures as much as possible. > > As maintainer of the LTTng out-of-tree kernel tracer, I appreciate the > effort to decouple instrumentation from the subsystem instrumentation, > but as long as the structure sits in public headers and the global > variables used within the TP_fast_assign() logic (e.g. init_user_ns) > are export-gpl, this is enough to make it easy for tracer integration Thank you for your friendly elaboration and suggestion! I will update this part based on your suggestion in next version. Thanks, Hongbo > and it keeps the tracepoint caller code footprint to a minimum. > > The TRACE_EVENT definitions are specific to the subsystem anyway, > so I don't think it matters that the TRACE_EVENT() need to access > the dentry and attr structures. > > So I agree with Steven's suggestion. However, just as a precision, > I suspect it will have mainly an impact on code size, but not > necessarily on icache footprint, because it will shrink the code > size within the tracepoint unlikely branch (cold instructions). > > Thanks, > > Mathieu > >> >> Thanks, >> Hongbo >> >>> >>> -- Steve >>> >
diff --git a/fs/hugetlbfs/inode.c b/fs/hugetlbfs/inode.c index 412f295acebe..f3399c6a02ca 100644 --- a/fs/hugetlbfs/inode.c +++ b/fs/hugetlbfs/inode.c @@ -39,6 +39,9 @@ #include <linux/uaccess.h> #include <linux/sched/mm.h> +#define CREATE_TRACE_POINTS +#include <trace/events/hugetlbfs.h> + static const struct address_space_operations hugetlbfs_aops; static const struct file_operations hugetlbfs_file_operations; static const struct inode_operations hugetlbfs_dir_inode_operations; @@ -686,6 +689,7 @@ static void hugetlbfs_evict_inode(struct inode *inode) { struct resv_map *resv_map; + trace_hugetlbfs_evict_inode(inode); remove_inode_hugepages(inode, 0, LLONG_MAX); /* @@ -813,8 +817,10 @@ static long hugetlbfs_fallocate(struct file *file, int mode, loff_t offset, if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE)) return -EOPNOTSUPP; - if (mode & FALLOC_FL_PUNCH_HOLE) - return hugetlbfs_punch_hole(inode, offset, len); + if (mode & FALLOC_FL_PUNCH_HOLE) { + error = hugetlbfs_punch_hole(inode, offset, len); + goto out_nolock; + } /* * Default preallocate case. @@ -918,6 +924,9 @@ static long hugetlbfs_fallocate(struct file *file, int mode, loff_t offset, inode_set_ctime_current(inode); out: inode_unlock(inode); + +out_nolock: + trace_hugetlbfs_fallocate(inode, mode, offset, len, error); return error; } @@ -934,6 +943,12 @@ static int hugetlbfs_setattr(struct mnt_idmap *idmap, if (error) return error; + trace_hugetlbfs_setattr(inode, dentry->d_name.len, dentry->d_name.name, + attr->ia_valid, attr->ia_mode, + from_kuid(&init_user_ns, attr->ia_uid), + from_kgid(&init_user_ns, attr->ia_gid), + inode->i_size, attr->ia_size); + if (ia_valid & ATTR_SIZE) { loff_t oldsize = inode->i_size; loff_t newsize = attr->ia_size; @@ -1032,6 +1047,7 @@ static struct inode *hugetlbfs_get_inode(struct super_block *sb, break; } lockdep_annotate_inode_mutex_key(inode); + trace_hugetlbfs_alloc_inode(inode, dir, mode); } else { if (resv_map) kref_put(&resv_map->refs, resv_map_release); @@ -1274,6 +1290,7 @@ static struct inode *hugetlbfs_alloc_inode(struct super_block *sb) static void hugetlbfs_free_inode(struct inode *inode) { + trace_hugetlbfs_free_inode(inode); kmem_cache_free(hugetlbfs_inode_cachep, HUGETLBFS_I(inode)); }
Here we use the hugetlbfs tracepoint to track the call stack. And the output in trace is as follows: ``` touch-5307 [004] ..... 1402.167607: hugetlbfs_alloc_inode: dev = (0,50), ino = 21380, dir = 16921, mode = 0100644 touch-5307 [004] ..... 1402.167638: hugetlbfs_setattr: dev = (0,50), ino = 21380, name = testfile1, ia_valid = 131184, ia_mode = 0132434, ia_uid = 2863018275, ia_gid = 4294967295, old_size = 0, ia_size = 4064 truncate-5328 [003] ..... 1436.031054: hugetlbfs_setattr: dev = (0,50), ino = 21380, name = testfile1, ia_valid = 8296, ia_mode = 0177777, ia_uid = 2862574544, ia_gid = 4294967295, old_size = 0, ia_size = 2097152 rm-5338 [004] ..... 1484.426247: hugetlbfs_evict_inode: dev = (0,50), ino = 21380, i_mode = 0100644, i_size = 2097152, i_nlink = 0, seals = 1, i_blocks = 0 <idle>-0 [004] ..s1. 1484.446668: hugetlbfs_free_inode: dev = (0,50), ino = 21380, i_mode = 0100644, i_size = 2097152, i_nlink = 0, seals = 1, i_blocks = 0 ``` Signed-off-by: Hongbo Li <lihongbo22@huawei.com> --- fs/hugetlbfs/inode.c | 21 +++++++++++++++++++-- 1 file changed, 19 insertions(+), 2 deletions(-)