Message ID | 1336748577-9258-1-git-send-email-saugata.das@stericsson.com |
---|---|
State | Superseded, archived |
Headers | show |
Hi, On Fri, 11 May 2012 20:32:56 +0530, Saugata Das wrote: > From: Saugata Das <saugata.das@linaro.org> > > Today, storage devices like eMMC has special features like data tagging > (introduced in MMC-4.5 version) in order to improve performance of some > specific writes. On MMC stack, data tagging is used for all writes which > has REQ_META flag set. This patch adds the capability to add REQ_META flag > during meta data write. > AFAIK, the REQ_META is only for marking a bio/req to be recognized when using blktrace or something. You can use REQ_PRIO for the purpose but it applies only if your ioscheduler is CFQ. But I'm not aware how the MMC stack works, so I might be missing something. Thanks, Namhyung > Signed-off-by: Saugata Das <saugata.das@linaro.org> > --- > fs/buffer.c | 10 ++++++++-- > include/linux/buffer_head.h | 2 ++ > 2 files changed, 10 insertions(+), 2 deletions(-) > > diff --git a/fs/buffer.c b/fs/buffer.c > index 36d6665..688b38b 100644 > --- a/fs/buffer.c > +++ b/fs/buffer.c > @@ -1685,7 +1685,10 @@ static int __block_write_full_page(struct inode *inode, struct page *page, > do { > struct buffer_head *next = bh->b_this_page; > if (buffer_async_write(bh)) { > - submit_bh(write_op, bh); > + if (buffer_meta(bh)) > + submit_bh(write_op | REQ_META, bh); > + else > + submit_bh(write_op, bh); > nr_underway++; > } > bh = next; > @@ -1739,7 +1742,10 @@ recover: > struct buffer_head *next = bh->b_this_page; > if (buffer_async_write(bh)) { > clear_buffer_dirty(bh); > - submit_bh(write_op, bh); > + if (buffer_meta(bh)) > + submit_bh(write_op | REQ_META, bh); > + else > + submit_bh(write_op, bh); > nr_underway++; > } > bh = next; > diff --git a/include/linux/buffer_head.h b/include/linux/buffer_head.h > index ef26043..0776564 100644 > --- a/include/linux/buffer_head.h > +++ b/include/linux/buffer_head.h > @@ -34,6 +34,7 @@ enum bh_state_bits { > BH_Write_EIO, /* I/O error on write */ > BH_Unwritten, /* Buffer is allocated on disk but not written */ > BH_Quiet, /* Buffer Error Prinks to be quiet */ > + BH_Meta, /* Is meta */ > > BH_PrivateStart,/* not a state bit, but the first bit available > * for private allocation by other entities > @@ -126,6 +127,7 @@ BUFFER_FNS(Delay, delay) > BUFFER_FNS(Boundary, boundary) > BUFFER_FNS(Write_EIO, write_io_error) > BUFFER_FNS(Unwritten, unwritten) > +BUFFER_FNS(Meta, meta) > > #define bh_offset(bh) ((unsigned long)(bh)->b_data & ~PAGE_MASK) > #define touch_buffer(bh) mark_page_accessed(bh->b_page) -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 14 May 2012 07:12, Namhyung Kim <namhyung.kim@lge.com> wrote: > Hi, > > On Fri, 11 May 2012 20:32:56 +0530, Saugata Das wrote: >> From: Saugata Das <saugata.das@linaro.org> >> >> Today, storage devices like eMMC has special features like data tagging >> (introduced in MMC-4.5 version) in order to improve performance of some >> specific writes. On MMC stack, data tagging is used for all writes which >> has REQ_META flag set. This patch adds the capability to add REQ_META flag >> during meta data write. >> > > AFAIK, the REQ_META is only for marking a bio/req to be recognized when > using blktrace or something. You can use REQ_PRIO for the purpose but it > applies only if your ioscheduler is CFQ. > > But I'm not aware how the MMC stack works, so I might be missing something. > Today on ext4, REQ_META or REQ_PRIO are only used during read operation. For meta-data writes, no special flag (REQ_META or REQ_PRIO) is set. On eMMC, we depend on REQ_META flag to implement the "reliable write" and "data tag" feature, which are linked with additional reliability and performance for file system meta-data writes. > Thanks, > Namhyung > > >> Signed-off-by: Saugata Das <saugata.das@linaro.org> >> --- >> fs/buffer.c | 10 ++++++++-- >> include/linux/buffer_head.h | 2 ++ >> 2 files changed, 10 insertions(+), 2 deletions(-) >> >> diff --git a/fs/buffer.c b/fs/buffer.c >> index 36d6665..688b38b 100644 >> --- a/fs/buffer.c >> +++ b/fs/buffer.c >> @@ -1685,7 +1685,10 @@ static int __block_write_full_page(struct inode *inode, struct page *page, >> do { >> struct buffer_head *next = bh->b_this_page; >> if (buffer_async_write(bh)) { >> - submit_bh(write_op, bh); >> + if (buffer_meta(bh)) >> + submit_bh(write_op | REQ_META, bh); >> + else >> + submit_bh(write_op, bh); >> nr_underway++; >> } >> bh = next; >> @@ -1739,7 +1742,10 @@ recover: >> struct buffer_head *next = bh->b_this_page; >> if (buffer_async_write(bh)) { >> clear_buffer_dirty(bh); >> - submit_bh(write_op, bh); >> + if (buffer_meta(bh)) >> + submit_bh(write_op | REQ_META, bh); >> + else >> + submit_bh(write_op, bh); >> nr_underway++; >> } >> bh = next; >> diff --git a/include/linux/buffer_head.h b/include/linux/buffer_head.h >> index ef26043..0776564 100644 >> --- a/include/linux/buffer_head.h >> +++ b/include/linux/buffer_head.h >> @@ -34,6 +34,7 @@ enum bh_state_bits { >> BH_Write_EIO, /* I/O error on write */ >> BH_Unwritten, /* Buffer is allocated on disk but not written */ >> BH_Quiet, /* Buffer Error Prinks to be quiet */ >> + BH_Meta, /* Is meta */ >> >> BH_PrivateStart,/* not a state bit, but the first bit available >> * for private allocation by other entities >> @@ -126,6 +127,7 @@ BUFFER_FNS(Delay, delay) >> BUFFER_FNS(Boundary, boundary) >> BUFFER_FNS(Write_EIO, write_io_error) >> BUFFER_FNS(Unwritten, unwritten) >> +BUFFER_FNS(Meta, meta) >> >> #define bh_offset(bh) ((unsigned long)(bh)->b_data & ~PAGE_MASK) >> #define touch_buffer(bh) mark_page_accessed(bh->b_page) -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Hi Ted, Christoph Will you please comment on the following patch set, [PATCH 1/2] block: add BH_Meta flag [PATCH 2/2] ext4: annotate all meta data requests If there is no remark, then will you please merge them to the next version. Thanks, Saugata On 11 May 2012 20:32, Saugata Das <saugata.das@stericsson.com> wrote: > From: Saugata Das <saugata.das@linaro.org> > > Today, storage devices like eMMC has special features like data tagging > (introduced in MMC-4.5 version) in order to improve performance of some > specific writes. On MMC stack, data tagging is used for all writes which > has REQ_META flag set. This patch adds the capability to add REQ_META flag > during meta data write. > > Signed-off-by: Saugata Das <saugata.das@linaro.org> > --- > fs/buffer.c | 10 ++++++++-- > include/linux/buffer_head.h | 2 ++ > 2 files changed, 10 insertions(+), 2 deletions(-) > > diff --git a/fs/buffer.c b/fs/buffer.c > index 36d6665..688b38b 100644 > --- a/fs/buffer.c > +++ b/fs/buffer.c > @@ -1685,7 +1685,10 @@ static int __block_write_full_page(struct inode *inode, struct page *page, > do { > struct buffer_head *next = bh->b_this_page; > if (buffer_async_write(bh)) { > - submit_bh(write_op, bh); > + if (buffer_meta(bh)) > + submit_bh(write_op | REQ_META, bh); > + else > + submit_bh(write_op, bh); > nr_underway++; > } > bh = next; > @@ -1739,7 +1742,10 @@ recover: > struct buffer_head *next = bh->b_this_page; > if (buffer_async_write(bh)) { > clear_buffer_dirty(bh); > - submit_bh(write_op, bh); > + if (buffer_meta(bh)) > + submit_bh(write_op | REQ_META, bh); > + else > + submit_bh(write_op, bh); > nr_underway++; > } > bh = next; > diff --git a/include/linux/buffer_head.h b/include/linux/buffer_head.h > index ef26043..0776564 100644 > --- a/include/linux/buffer_head.h > +++ b/include/linux/buffer_head.h > @@ -34,6 +34,7 @@ enum bh_state_bits { > BH_Write_EIO, /* I/O error on write */ > BH_Unwritten, /* Buffer is allocated on disk but not written */ > BH_Quiet, /* Buffer Error Prinks to be quiet */ > + BH_Meta, /* Is meta */ > > BH_PrivateStart,/* not a state bit, but the first bit available > * for private allocation by other entities > @@ -126,6 +127,7 @@ BUFFER_FNS(Delay, delay) > BUFFER_FNS(Boundary, boundary) > BUFFER_FNS(Write_EIO, write_io_error) > BUFFER_FNS(Unwritten, unwritten) > +BUFFER_FNS(Meta, meta) > > #define bh_offset(bh) ((unsigned long)(bh)->b_data & ~PAGE_MASK) > #define touch_buffer(bh) mark_page_accessed(bh->b_page) > -- > 1.7.4.3 > -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 05/11/2012 06:02 PM, Saugata Das wrote: > From: Saugata Das <saugata.das@linaro.org> > > Today, storage devices like eMMC has special features like data tagging > (introduced in MMC-4.5 version) in order to improve performance of some > specific writes. On MMC stack, data tagging is used for all writes which > has REQ_META flag set. This patch adds the capability to add REQ_META flag > during meta data write. > > Signed-off-by: Saugata Das <saugata.das@linaro.org> > --- > fs/buffer.c | 10 ++++++++-- > include/linux/buffer_head.h | 2 ++ > 2 files changed, 10 insertions(+), 2 deletions(-) > > diff --git a/fs/buffer.c b/fs/buffer.c > index 36d6665..688b38b 100644 > --- a/fs/buffer.c > +++ b/fs/buffer.c > @@ -1685,7 +1685,10 @@ static int __block_write_full_page(struct inode *inode, struct page *page, > do { > struct buffer_head *next = bh->b_this_page; > if (buffer_async_write(bh)) { > - submit_bh(write_op, bh); > + if (buffer_meta(bh)) > + submit_bh(write_op | REQ_META, bh); > + else > + submit_bh(write_op, bh); Its not nice to duplicate a call site for the parameter difference it's better to change the parameter and call the function in one place. (an assembler call site can get big) You can do: + submit_bh(write_op | (buffer_meta(bh) << __REQ_META), bh); And also avoid a conditional inside a loop. > nr_underway++; > } > bh = next; > @@ -1739,7 +1742,10 @@ recover: > struct buffer_head *next = bh->b_this_page; > if (buffer_async_write(bh)) { > clear_buffer_dirty(bh); > - submit_bh(write_op, bh); > + if (buffer_meta(bh)) > + submit_bh(write_op | REQ_META, bh); > + else > + submit_bh(write_op, bh); Here too Boaz > nr_underway++; > } > bh = next; > diff --git a/include/linux/buffer_head.h b/include/linux/buffer_head.h > index ef26043..0776564 100644 > --- a/include/linux/buffer_head.h > +++ b/include/linux/buffer_head.h > @@ -34,6 +34,7 @@ enum bh_state_bits { > BH_Write_EIO, /* I/O error on write */ > BH_Unwritten, /* Buffer is allocated on disk but not written */ > BH_Quiet, /* Buffer Error Prinks to be quiet */ > + BH_Meta, /* Is meta */ > > BH_PrivateStart,/* not a state bit, but the first bit available > * for private allocation by other entities > @@ -126,6 +127,7 @@ BUFFER_FNS(Delay, delay) > BUFFER_FNS(Boundary, boundary) > BUFFER_FNS(Write_EIO, write_io_error) > BUFFER_FNS(Unwritten, unwritten) > +BUFFER_FNS(Meta, meta) > > #define bh_offset(bh) ((unsigned long)(bh)->b_data & ~PAGE_MASK) > #define touch_buffer(bh) mark_page_accessed(bh->b_page) -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 16 May 2012 18:02, Boaz Harrosh <bharrosh@panasas.com> wrote: > On 05/11/2012 06:02 PM, Saugata Das wrote: > >> From: Saugata Das <saugata.das@linaro.org> >> >> Today, storage devices like eMMC has special features like data tagging >> (introduced in MMC-4.5 version) in order to improve performance of some >> specific writes. On MMC stack, data tagging is used for all writes which >> has REQ_META flag set. This patch adds the capability to add REQ_META flag >> during meta data write. >> >> Signed-off-by: Saugata Das <saugata.das@linaro.org> >> --- >> fs/buffer.c | 10 ++++++++-- >> include/linux/buffer_head.h | 2 ++ >> 2 files changed, 10 insertions(+), 2 deletions(-) >> >> diff --git a/fs/buffer.c b/fs/buffer.c >> index 36d6665..688b38b 100644 >> --- a/fs/buffer.c >> +++ b/fs/buffer.c >> @@ -1685,7 +1685,10 @@ static int __block_write_full_page(struct inode *inode, struct page *page, >> do { >> struct buffer_head *next = bh->b_this_page; >> if (buffer_async_write(bh)) { >> - submit_bh(write_op, bh); >> + if (buffer_meta(bh)) >> + submit_bh(write_op | REQ_META, bh); >> + else >> + submit_bh(write_op, bh); > > > Its not nice to duplicate a call site for the parameter difference > it's better to change the parameter and call the function in one > place. (an assembler call site can get big) > > You can do: > + submit_bh(write_op | (buffer_meta(bh) << __REQ_META), bh); > > And also avoid a conditional inside a loop. > Thanks for your comments. I will take care of this in the next version. > >> nr_underway++; >> } >> bh = next; >> @@ -1739,7 +1742,10 @@ recover: >> struct buffer_head *next = bh->b_this_page; >> if (buffer_async_write(bh)) { >> clear_buffer_dirty(bh); >> - submit_bh(write_op, bh); >> + if (buffer_meta(bh)) >> + submit_bh(write_op | REQ_META, bh); >> + else >> + submit_bh(write_op, bh); > > > Here too > > Boaz > >> nr_underway++; >> } >> bh = next; >> diff --git a/include/linux/buffer_head.h b/include/linux/buffer_head.h >> index ef26043..0776564 100644 >> --- a/include/linux/buffer_head.h >> +++ b/include/linux/buffer_head.h >> @@ -34,6 +34,7 @@ enum bh_state_bits { >> BH_Write_EIO, /* I/O error on write */ >> BH_Unwritten, /* Buffer is allocated on disk but not written */ >> BH_Quiet, /* Buffer Error Prinks to be quiet */ >> + BH_Meta, /* Is meta */ >> >> BH_PrivateStart,/* not a state bit, but the first bit available >> * for private allocation by other entities >> @@ -126,6 +127,7 @@ BUFFER_FNS(Delay, delay) >> BUFFER_FNS(Boundary, boundary) >> BUFFER_FNS(Write_EIO, write_io_error) >> BUFFER_FNS(Unwritten, unwritten) >> +BUFFER_FNS(Meta, meta) >> >> #define bh_offset(bh) ((unsigned long)(bh)->b_data & ~PAGE_MASK) >> #define touch_buffer(bh) mark_page_accessed(bh->b_page) > > -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/fs/buffer.c b/fs/buffer.c index 36d6665..688b38b 100644 --- a/fs/buffer.c +++ b/fs/buffer.c @@ -1685,7 +1685,10 @@ static int __block_write_full_page(struct inode *inode, struct page *page, do { struct buffer_head *next = bh->b_this_page; if (buffer_async_write(bh)) { - submit_bh(write_op, bh); + if (buffer_meta(bh)) + submit_bh(write_op | REQ_META, bh); + else + submit_bh(write_op, bh); nr_underway++; } bh = next; @@ -1739,7 +1742,10 @@ recover: struct buffer_head *next = bh->b_this_page; if (buffer_async_write(bh)) { clear_buffer_dirty(bh); - submit_bh(write_op, bh); + if (buffer_meta(bh)) + submit_bh(write_op | REQ_META, bh); + else + submit_bh(write_op, bh); nr_underway++; } bh = next; diff --git a/include/linux/buffer_head.h b/include/linux/buffer_head.h index ef26043..0776564 100644 --- a/include/linux/buffer_head.h +++ b/include/linux/buffer_head.h @@ -34,6 +34,7 @@ enum bh_state_bits { BH_Write_EIO, /* I/O error on write */ BH_Unwritten, /* Buffer is allocated on disk but not written */ BH_Quiet, /* Buffer Error Prinks to be quiet */ + BH_Meta, /* Is meta */ BH_PrivateStart,/* not a state bit, but the first bit available * for private allocation by other entities @@ -126,6 +127,7 @@ BUFFER_FNS(Delay, delay) BUFFER_FNS(Boundary, boundary) BUFFER_FNS(Write_EIO, write_io_error) BUFFER_FNS(Unwritten, unwritten) +BUFFER_FNS(Meta, meta) #define bh_offset(bh) ((unsigned long)(bh)->b_data & ~PAGE_MASK) #define touch_buffer(bh) mark_page_accessed(bh->b_page)