Skip to contentSkip to conversation

mkfs: inform during block discarding

PATCH Superseded Pavel Reichl 4 participants 22 messages Nov 21, 2019 – Nov 26, 2019

v1 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

cover mkfs: inform during block discarding

1st patch breaks discarding into smaller chunks and thus make interruption and logging possible.

2nd patch writes messages about discarding process into stderr. Sample output looks like this:
Discarding: 0% done
Discarding: 20% done
Discarding: 40% done
Discarding: 60% done
Discarding: 80% done
Discarding is done.

Pavel Reichl (2):
mkfs: Break block discard into chunks of 2 GB
mkfs: Show progress during block discard

 mkfs/xfs_mkfs.c | 40 +++++++++++++++++++++++++++++++++-------
 1 file changed, 33 insertions(+), 7 deletions(-)
Signature · 1 line
2.23.0
#1 lore raw

PATCH 1/2 mkfs: Break block discard into chunks of 2 GB

---
 mkfs/xfs_mkfs.c | 32 +++++++++++++++++++++++++-------
 1 file changed, 25 insertions(+), 7 deletions(-)
mkfs/xfs_mkfs.c +25 -7
index 18338a61..a02d6f66 100644
@@ -1242,15 +1242,33 @@ done:
 static void
 discard_blocks(dev_t dev, uint64_t nsectors)
 {
-	int fd;
+	int		fd;
+	uint64_t	offset		= 0;
+	/* Maximal chunk of bytes to discard is 2GB */
+	const uint64_t	step		= (uint64_t)2<<30;
+	/* Sector size is 512 bytes */
+	const uint64_t	count		= nsectors << 9;
 
-	/*
-	 * We intentionally ignore errors from the discard ioctl.  It is
-	 * not necessary for the mkfs functionality but just an optimization.
-	 */
 	fd = libxfs_device_to_fd(dev);
-	if (fd > 0)
-		platform_discard_blocks(fd, 0, nsectors << 9);
+	if (fd <= 0)
+		return;
+
+	while (offset < count) {
+		uint64_t	tmp_step = step;
+
+		if ((offset + step) > count)
+			tmp_step = count - offset;
+
+		/*
+		 * We intentionally ignore errors from the discard ioctl. It is
+		 * not necessary for the mkfs functionality but just an
+		 * optimization. However we should stop on error.
+		 */
+		if (platform_discard_blocks(fd, offset, tmp_step))
+			return;
+
+		offset += tmp_step;
+	}
 }
 
 static __attribute__((noreturn)) void
Signature · 1 line
2.23.0
#2 lore raw

Re: [PATCH 1/2] mkfs: Break block discard into chunks of 2 GB

Quoted 10 lines · Pavel Reichl
---
 mkfs/xfs_mkfs.c | 32 +++++++++++++++++++++++++-------
 1 file changed, 25 insertions(+), 7 deletions(-)
mkfs/xfs_mkfs.c +0 -0
index 18338a61..a02d6f66 100644

mkfs/xfs_mkfs.c done: · line 1248

Quoted 8 lines
mkfs/xfs_mkfs.c +4 -1
@@ -1242,15 +1242,33 @@ done:
 static void
 discard_blocks(dev_t dev, uint64_t nsectors)
 {
-	int fd;
+	int		fd;
+	uint64_t	offset		= 0;
+	/* Maximal chunk of bytes to discard is 2GB */
+	const uint64_t	step		= (uint64_t)2<<30;

You don't need the tabs after the variable name, e.g.

	/* Maximal chunk of bytes to discard is 2GB */
	const uint64_t	step = 2ULL << 30;

mkfs/xfs_mkfs.c done: · line 1250

Quoted 2 lines
mkfs/xfs_mkfs.c +2 -0
+	/* Sector size is 512 bytes */
+	const uint64_t	count		= nsectors << 9;

count = BBTOB(nsectors)?

mkfs/xfs_mkfs.c done: · line 1257

Quoted 13 lines
mkfs/xfs_mkfs.c +5 -6
-	/*
-	 * We intentionally ignore errors from the discard ioctl.  It is
-	 * not necessary for the mkfs functionality but just an optimization.
-	 */
 	fd = libxfs_device_to_fd(dev);
-	if (fd > 0)
-		platform_discard_blocks(fd, 0, nsectors << 9);
+	if (fd <= 0)
+		return;
+
+	while (offset < count) {
+		uint64_t	tmp_step = step;

tmp_step = min(step, count - offset); ?

Otherwise seems reasonable to me, if nothing else to avoid the problem where you ask mkfs to discard and can't cancel it....

--D

Quoted 20 lines
mkfs/xfs_mkfs.c +14 -0
+
+		if ((offset + step) > count)
+			tmp_step = count - offset;
+
+		/*
+		 * We intentionally ignore errors from the discard ioctl. It is
+		 * not necessary for the mkfs functionality but just an
+		 * optimization. However we should stop on error.
+		 */
+		if (platform_discard_blocks(fd, offset, tmp_step))
+			return;
+
+		offset += tmp_step;
+	}
 }
 
 static __attribute__((noreturn)) void
Signature · 1 line
2.23.0
#3 lore raw

Re: [PATCH 1/2] mkfs: Break block discard into chunks of 2 GB

Thanks Darrick for the comments. It makes sense to me, the next iteration of the patchset will address that.

Quoted 73 lines · Darrick J. Wong
Quoted 10 lines · Pavel Reichl
---
 mkfs/xfs_mkfs.c | 32 +++++++++++++++++++++++++-------
 1 file changed, 25 insertions(+), 7 deletions(-)
mkfs/xfs_mkfs.c +0 -0
index 18338a61..a02d6f66 100644

mkfs/xfs_mkfs.c done: · line 1248

Quoted 8 lines
mkfs/xfs_mkfs.c +4 -1
@@ -1242,15 +1242,33 @@ done:
 static void
 discard_blocks(dev_t dev, uint64_t nsectors)
 {
-     int fd;
+     int             fd;
+     uint64_t        offset          = 0;
+     /* Maximal chunk of bytes to discard is 2GB */
+     const uint64_t  step            = (uint64_t)2<<30;

You don't need the tabs after the variable name, e.g.

        /* Maximal chunk of bytes to discard is 2GB */
        const uint64_t  step = 2ULL << 30;

mkfs/xfs_mkfs.c done: · line 1250

Quoted 2 lines
mkfs/xfs_mkfs.c +2 -0
+     /* Sector size is 512 bytes */
+     const uint64_t  count           = nsectors << 9;

count = BBTOB(nsectors)?

mkfs/xfs_mkfs.c done: · line 1257

Quoted 13 lines
mkfs/xfs_mkfs.c +5 -6
-     /*
-      * We intentionally ignore errors from the discard ioctl.  It is
-      * not necessary for the mkfs functionality but just an optimization.
-      */
      fd = libxfs_device_to_fd(dev);
-     if (fd > 0)
-             platform_discard_blocks(fd, 0, nsectors << 9);
+     if (fd <= 0)
+             return;
+
+     while (offset < count) {
+             uint64_t        tmp_step = step;

tmp_step = min(step, count - offset); ?

Otherwise seems reasonable to me, if nothing else to avoid the problem where you ask mkfs to discard and can't cancel it....

--D

Quoted 20 lines
mkfs/xfs_mkfs.c +14 -1
+
+             if ((offset + step) > count)
+                     tmp_step = count - offset;
+
+             /*
+              * We intentionally ignore errors from the discard ioctl. It is
+              * not necessary for the mkfs functionality but just an
+              * optimization. However we should stop on error.
+              */
+             if (platform_discard_blocks(fd, offset, tmp_step))
+                     return;
+
+             offset += tmp_step;
+     }
 }
 
 static __attribute__((noreturn)) void
--

2.23.0

#4 lore raw

Re: [PATCH 1/2] mkfs: Break block discard into chunks of 2 GB

In reply to Darrick J. Wong
Quoted 1 line · Darrick J. Wong

On Thu, Nov 21, 2019 at 10:44:44PM +0100, Pavel Reichl wrote:

concur w/ others that a reason for the change (and a reason for the size selection) would be appropriate to have in the changelog.

Quoted 21 lines
Quoted 10 lines
---
 mkfs/xfs_mkfs.c | 32 +++++++++++++++++++++++++-------
 1 file changed, 25 insertions(+), 7 deletions(-)
mkfs/xfs_mkfs.c +0 -0
index 18338a61..a02d6f66 100644

mkfs/xfs_mkfs.c done: · line 1248

Quoted 8 lines
mkfs/xfs_mkfs.c +4 -1
@@ -1242,15 +1242,33 @@ done:
 static void
 discard_blocks(dev_t dev, uint64_t nsectors)
 {
-	int fd;
+	int		fd;
+	uint64_t	offset		= 0;
+	/* Maximal chunk of bytes to discard is 2GB */
+	const uint64_t	step		= (uint64_t)2<<30;

You don't need the tabs after the variable name, e.g.

Quoted 7 lines
	/* Maximal chunk of bytes to discard is 2GB */
	const uint64_t	step = 2ULL << 30;

mkfs/xfs_mkfs.c done: · line 1250

Quoted 2 lines
mkfs/xfs_mkfs.c +2 -0
+	/* Sector size is 512 bytes */
+	const uint64_t	count		= nsectors << 9;

count = BBTOB(nsectors)?

FYI this is a macro that xfs developers have learned about. ;) It stands for "Basic Block TO Byte" where "basic block" pretty much means "512-byte sector."

-Eric

#5 lore raw

Re: [PATCH 1/2] mkfs: Break block discard into chunks of 2 GB

In reply to Pavel Reichl
Quoted 2 lines · Pavel Reichl

---

This is mixing an explanation about why the change is being made and what was considered when making decisions about the change.

e.g. my first questions on looking at the patch were:

- why do we need to break up the discards into 2GB chunks?
- why 2GB?
- why not use libblkid to query the maximum discard size and use that as the step size instead?
- is there any performance impact from breaking up large discards that might be optimised by the kernel into many overlapping async operations into small, synchronous discards?

i.e. the reviewer can read what the patch does, but that deosn't explain why the patch does this. Hence it's a good idea to explain the problem being solved or the feature requirements that have lead to the changes in the patch....

Cheers,

Dave.

Signature · 2 lines
Dave Chinner
[email protected]
#6 lore raw

Re: [PATCH 1/2] mkfs: Break block discard into chunks of 2 GB

Quoted 11 lines · Dave Chinner
Quoted 2 lines · Pavel Reichl

---

This is mixing an explanation about why the change is being made and what was considered when making decisions about the change.

e.g. my first questions on looking at the patch were:

- why do we need to break up the discards into 2GB chunks?
- why 2GB?

Yeah, I'm wondering that too.

Quoted 2 lines

- why not use libblkid to query the maximum discard size and use that as the step size instead?

FWIW my SATA SSDs the discard-max is 2G whereas on the NVME it's 2T. I guess firmwares have gotten 1000x better in the past few years, possibly because of the hundred or so 10x programmers that they've all been hiring.

Quoted 4 lines

- is there any performance impact from breaking up large discards that might be optimised by the kernel into many overlapping async operations into small, synchronous discards?

Also:
What is the end goal that you have in mind? Is the progress reporting the ultimate goal? Or is it to break up the BLKDISCARD calls so that someone can ^C a mkfs operation and not have it just sit there continuing to run?

--D

Quoted 11 lines

i.e. the reviewer can read what the patch does, but that deosn't explain why the patch does this. Hence it's a good idea to explain the problem being solved or the feature requirements that have lead to the changes in the patch....

Cheers,

Dave.

Signature · 2 lines
Dave Chinner
[email protected]
#7 lore raw

Re: [PATCH 1/2] mkfs: Break block discard into chunks of 2 GB

Quoted 9 lines · Darrick J. Wong
Quoted 7 lines · Dave Chinner
Quoted 2 lines · Pavel Reichl

---

This is mixing an explanation about why the change is being made and what was considered when making decisions about the change.

Quoted 6 lines
Quoted 4 lines

e.g. my first questions on looking at the patch were:

- why do we need to break up the discards into 2GB chunks?
- why 2GB?

Yeah, I'm wondering that too.

OK, thank you both for the question - simple answer is that I took what is used in e2fsprogs as default and I expected a discussion about proper value during review process :-)

Quoted 13 lines
Quoted 3 lines

- why not use libblkid to query the maximum discard size and use that as the step size instead?

FWIW my SATA SSDs the discard-max is 2G whereas on the NVME it's 2T. I guess firmwares have gotten 1000x better in the past few years, possibly because of the hundred or so 10x programmers that they've all been hiring.

Quoted 4 lines

- is there any performance impact from breaking up large discards that might be optimised by the kernel into many overlapping async operations into small, synchronous discards?

Quoted 5 lines

Also:
What is the end goal that you have in mind? Is the progress reporting the ultimate goal? Or is it to break up the BLKDISCARD calls so that someone can ^C a mkfs operation and not have it just sit there continuing to run?

The goal is mainly the progress reporting but the possibility to do ^C is also convenient. It seems that some users are not happy about the BLKDISCARD taking too long and at the same time not being informed about that - so they think that the command actually hung.

Quoted 15 lines

--D

Quoted 11 lines

i.e. the reviewer can read what the patch does, but that deosn't explain why the patch does this. Hence it's a good idea to explain the problem being solved or the feature requirements that have lead to the changes in the patch....

Cheers,

Dave.
--
Dave Chinner
[email protected]

#8 lore raw

Re: [PATCH 1/2] mkfs: Break block discard into chunks of 2 GB

Quoted 11 lines · Pavel Reichl
Quoted 5 lines · Darrick J. Wong

Also:
What is the end goal that you have in mind? Is the progress reporting the ultimate goal? Or is it to break up the BLKDISCARD calls so that someone can ^C a mkfs operation and not have it just sit there continuing to run?

The goal is mainly the progress reporting but the possibility to do ^C is also convenient. It seems that some users are not happy about the BLKDISCARD taking too long and at the same time not being informed about that - so they think that the command actually hung.

Ok, that's a good summary to put in the commit description - it tells the reviewer exactly what you are trying to acheive, and gives them context to evaluate it against.

Cheers,

Dave.

Signature · 2 lines
Dave Chinner
[email protected]
#9 lore raw

Re: [PATCH 1/2] mkfs: Break block discard into chunks of 2 GB

In reply to Dave Chinner
Quoted 7 lines · Dave Chinner
Quoted 2 lines · Pavel Reichl

---

This is mixing an explanation about why the change is being made and what was considered when making decisions about the change.

Thanks, I'll try to improve that.

Quoted 7 lines

e.g. my first questions on looking at the patch were:

- why do we need to break up the discards into 2GB chunks?
- why 2GB?
- why not use libblkid to query the maximum discard size and use that as the step size instead?

This is new for me, please let me learn more about that.

Quoted 4 lines

- is there any performance impact from breaking up large discards that might be optimised by the kernel into many overlapping async operations into small, synchronous discards?

Honestly, I don't have an answer for that ATM - it's quite possible. It certainly needs more investigating. On the other hand - current lack of feedback causes user discomfort. So I'd like to know your opinion - should the change proposed by this patch be default behaviour (as it may be more user friendly) and should we add an option that would 'revert' to current behaviour (that would be for informed user).

Quoted 13 lines

i.e. the reviewer can read what the patch does, but that deosn't explain why the patch does this. Hence it's a good idea to explain the problem being solved or the feature requirements that have lead to the changes in the patch....

Cheers,

Dave.
--
Dave Chinner
[email protected]

#10 lore raw

Re: [PATCH 1/2] mkfs: Break block discard into chunks of 2 GB

In reply to Dave Chinner
Quoted 7 lines · Dave Chinner
Quoted 2 lines · Pavel Reichl

---

This is mixing an explanation about why the change is being made and what was considered when making decisions about the change.

Quoted 6 lines

e.g. my first questions on looking at the patch were:

- why do we need to break up the discards into 2GB chunks?
- why 2GB?
- why not use libblkid to query the maximum discard size and use that as the step size instead?

Just wondering, can we trust that to be reasonably performant? (the whole motivation here is for hardware that takes inordinately long to do discard, I wonder if we can count on such hardware to properly fill out this info....)

Quoted 4 lines

- is there any performance impact from breaking up large discards that might be optimised by the kernel into many overlapping async operations into small, synchronous discards?

FWIW, I had simply suggested to Pavel that he follow e2fsprogs' lead here - afaik they haven't had issues/complaints with their 2g iteration, and at one point Lukas did some investigation into the size selection...

Thanks,
-Eric

Quoted 9 lines

i.e. the reviewer can read what the patch does, but that deosn't explain why the patch does this. Hence it's a good idea to explain the problem being solved or the feature requirements that have lead to the changes in the patch....

Cheers,

Dave.

#11 lore raw

Re: [PATCH 1/2] mkfs: Break block discard into chunks of 2 GB

Quoted 15 lines · Eric Sandeen
Quoted 7 lines · Dave Chinner
Quoted 2 lines · Pavel Reichl

---

This is mixing an explanation about why the change is being made and what was considered when making decisions about the change.

Quoted 6 lines

e.g. my first questions on looking at the patch were:

- why do we need to break up the discards into 2GB chunks?
- why 2GB?
- why not use libblkid to query the maximum discard size and use that as the step size instead?

Quoted 4 lines

Just wondering, can we trust that to be reasonably performant? (the whole motivation here is for hardware that takes inordinately long to do discard, I wonder if we can count on such hardware to properly fill out this info....)

Looking at the docs in kernel/Documentation/block/queue-sysfs.rst:

discard_max_hw_bytes (RO)
-------------------------
Devices that support discard functionality may have internal limits on the number of bytes that can be trimmed or unmapped in a single operation. The discard_max_bytes parameter is set by the device driver to the maximum number of bytes that can be discarded in a single operation. Discard requests issued to the device must not exceed this limit. A discard_max_bytes value of 0 means that the device does not support discard functionality.

discard_max_bytes (RW)
----------------------
While discard_max_hw_bytes is the hardware limit for the device, this setting is the software limit. Some devices exhibit large latencies when large discards are issued, setting this value lower will make Linux issue smaller discards and potentially help reduce latencies induced by large discard operations.

it seems like a strong suggestion that the discard_max_hw_bytes value may still be problematic, and discard_max_bytes can be hand-tuned to something smaller if it's a problem. To me that indicates that discard_max_hw_bytes probably can't be trusted to be performant, and presumably discard_max_bytes won't be either in that case unless it's been hand-tuned by the admin?

-Eric

#12 lore raw

Re: [PATCH 1/2] mkfs: Break block discard into chunks of 2 GB

Quoted 40 lines · Eric Sandeen
Quoted 15 lines · Eric Sandeen
Quoted 7 lines · Dave Chinner
Quoted 2 lines · Pavel Reichl

---

This is mixing an explanation about why the change is being made and what was considered when making decisions about the change.

Quoted 6 lines

e.g. my first questions on looking at the patch were:

- why do we need to break up the discards into 2GB chunks?
- why 2GB?
- why not use libblkid to query the maximum discard size and use that as the step size instead?

Quoted 4 lines

Just wondering, can we trust that to be reasonably performant? (the whole motivation here is for hardware that takes inordinately long to do discard, I wonder if we can count on such hardware to properly fill out this info....)

Looking at the docs in kernel/Documentation/block/queue-sysfs.rst:

discard_max_hw_bytes (RO)
-------------------------
Devices that support discard functionality may have internal limits on the number of bytes that can be trimmed or unmapped in a single operation. The discard_max_bytes parameter is set by the device driver to the maximum number of bytes that can be discarded in a single operation. Discard requests issued to the device must not exceed this limit. A discard_max_bytes value of 0 means that the device does not support discard functionality.

discard_max_bytes (RW)
----------------------
While discard_max_hw_bytes is the hardware limit for the device, this setting is the software limit. Some devices exhibit large latencies when large discards are issued, setting this value lower will make Linux issue smaller discards and potentially help reduce latencies induced by large discard operations.

Quoted 5 lines

it seems like a strong suggestion that the discard_max_hw_bytes value may still be problematic, and discard_max_bytes can be hand-tuned to something smaller if it's a problem. To me that indicates that discard_max_hw_bytes probably can't be trusted to be performant, and presumably discard_max_bytes won't be either in that case unless it's been hand-tuned by the admin?

Lukas, Jeff Moyer reminded me that you did a lot of investigation into this behavior a while back. Can you shed light on this, particularly how you chose 2G as the discard granularity for mke2fs?

Thanks,
-Eric

#13 lore raw

Re: [PATCH 1/2] mkfs: Break block discard into chunks of 2 GB

In reply to Pavel Reichl
Quoted 10 lines · Pavel Reichl
---
 mkfs/xfs_mkfs.c | 32 +++++++++++++++++++++++++-------
 1 file changed, 25 insertions(+), 7 deletions(-)
mkfs/xfs_mkfs.c +0 -0
index 18338a61..a02d6f66 100644

mkfs/xfs_mkfs.c done: · line 1248

Quoted 8 lines
mkfs/xfs_mkfs.c +4 -1
@@ -1242,15 +1242,33 @@ done:
 static void
 discard_blocks(dev_t dev, uint64_t nsectors)
 {
-	int fd;
+	int		fd;
+	uint64_t	offset		= 0;
+	/* Maximal chunk of bytes to discard is 2GB */
+	const uint64_t	step		= (uint64_t)2<<30;

Regarding the discard step size, I would like to just see us keep 2G - I see problems with the alternate suggestions proposed in the threads on this patch review:

1) query block device for maximal discard size
-> block device folks I've talked to (Jeff Moyer in particular) stated
that many devices are known for putting a huge value in here, and then taking far, far too long to process that size request. In short, maximum size != fast.

2) discard one AG size at a time
-> this can be up to 1T, which puts us right back at our problem of large,
slow discards. And in particular, AG size has no relation at all to a device's discard behavior. (further complicating this, we don't have this geometry available anywhere in the current chain of calls to the discard ioctl.)

Lukas did an investigation of discard behaviors (though it was some time ago https://sourceforge.net/projects/test-discard/) and arrived at 2G as a reasonable size after testing many different devices - I've not seen any complaints from mke2fs users about problems doing discards in 2G chunks.

So I think just picking a fixed 2G size is the best plan for now.

(one nitpick, I'd fix the comment above to not say "Maximal" because that sounds like some hard limit imposed by something other than the code; I'd just say "Discard the device 2G at a time" or something like that.)

A comment above the loop explaining in more detail that we iterate in step sizes so that the utility can be interrupted would probably be helpful.

Thanks,
-Eric

#14 lore raw

PATCH 2/2 mkfs: Show progress during block discard

In reply to Pavel Reichl
---
 mkfs/xfs_mkfs.c | 8 ++++++++
 1 file changed, 8 insertions(+)
mkfs/xfs_mkfs.c +8 -0
index a02d6f66..07b8bd78 100644
@@ -1248,6 +1248,7 @@ discard_blocks(dev_t dev, uint64_t nsectors)
 	const uint64_t	step		= (uint64_t)2<<30;
 	/* Sector size is 512 bytes */
 	const uint64_t	count		= nsectors << 9;
+	uint64_t	prev_done	= (uint64_t) ~0;
 
 	fd = libxfs_device_to_fd(dev);
 	if (fd <= 0)
@@ -1255,6 +1256,7 @@ discard_blocks(dev_t dev, uint64_t nsectors)
 
 	while (offset < count) {
 		uint64_t	tmp_step = step;
+		uint64_t	done = offset * 100 / count;
 
 		if ((offset + step) > count)
 			tmp_step = count - offset;
@@ -1268,7 +1270,13 @@ discard_blocks(dev_t dev, uint64_t nsectors)
 			return;
 
 		offset += tmp_step;
+
+		if (prev_done != done) {
+			prev_done = done;
+			fprintf(stderr, _("Discarding: %2lu%% done\n"), done);
+		}
 	}
+	fprintf(stderr, _("Discarding is done.\n"));
 }
 
 static __attribute__((noreturn)) void
Signature · 1 line
2.23.0
#15 lore raw

Re: [PATCH 2/2] mkfs: Show progress during block discard

Quoted 26 lines · Pavel Reichl
---
 mkfs/xfs_mkfs.c | 8 ++++++++
 1 file changed, 8 insertions(+)
mkfs/xfs_mkfs.c +2 -0
index a02d6f66..07b8bd78 100644
@@ -1248,6 +1248,7 @@ discard_blocks(dev_t dev, uint64_t nsectors)
 	const uint64_t	step		= (uint64_t)2<<30;
 	/* Sector size is 512 bytes */
 	const uint64_t	count		= nsectors << 9;
+	uint64_t	prev_done	= (uint64_t) ~0;
 
 	fd = libxfs_device_to_fd(dev);
 	if (fd <= 0)
@@ -1255,6 +1256,7 @@ discard_blocks(dev_t dev, uint64_t nsectors)
 
 	while (offset < count) {
 		uint64_t	tmp_step = step;
+		uint64_t	done = offset * 100 / count;
 
 		if ((offset + step) > count)
 			tmp_step = count - offset;

mkfs/xfs_mkfs.c discard_blocks(dev_t dev, uint64_t nsectors) · line 1274

Quoted 5 lines
mkfs/xfs_mkfs.c +2 -0
@@ -1268,7 +1270,13 @@ discard_blocks(dev_t dev, uint64_t nsectors)
 			return;
 
 		offset += tmp_step;
+
+		if (prev_done != done) {

Hmm... so this prints the status message every increase percentage point, right?

mkfs/xfs_mkfs.c discard_blocks(dev_t dev, uint64_t nsectors) · line 1276

Quoted 2 lines
mkfs/xfs_mkfs.c +2 -0
+			prev_done = done;
+			fprintf(stderr, _("Discarding: %2lu%% done\n"), done);

This isn't an error, so why output to stderr?

FWIW if it's a tty you might consider ending that string with \r so the status messages don't scroll off the screen. Or possibly only reporting status if stdout is a tty?

--D

Quoted 9 lines
mkfs/xfs_mkfs.c +2 -0
+		}
 	}
+	fprintf(stderr, _("Discarding is done.\n"));
 }
 
 static __attribute__((noreturn)) void
Signature · 1 line
2.23.0
#16 lore raw

Re: [PATCH 2/2] mkfs: Show progress during block discard

Quoted 30 lines · Darrick J. Wong
Quoted 28 lines · Pavel Reichl
---
 mkfs/xfs_mkfs.c | 8 ++++++++
 1 file changed, 8 insertions(+)
mkfs/xfs_mkfs.c +2 -0
index a02d6f66..07b8bd78 100644
@@ -1248,6 +1248,7 @@ discard_blocks(dev_t dev, uint64_t nsectors)
      const uint64_t  step            = (uint64_t)2<<30;
      /* Sector size is 512 bytes */
      const uint64_t  count           = nsectors << 9;
+     uint64_t        prev_done       = (uint64_t) ~0;
 
      fd = libxfs_device_to_fd(dev);
      if (fd <= 0)
@@ -1255,6 +1256,7 @@ discard_blocks(dev_t dev, uint64_t nsectors)
 
      while (offset < count) {
              uint64_t        tmp_step = step;
+             uint64_t        done = offset * 100 / count;
 
              if ((offset + step) > count)
                      tmp_step = count - offset;
@@ -1268,7 +1270,13 @@ discard_blocks(dev_t dev, uint64_t nsectors)
                      return;
 
Quoted 6 lines

mkfs/xfs_mkfs.c discard_blocks(dev_t dev, uint64_t nsectors) · line 1274

Quoted 3 lines
mkfs/xfs_mkfs.c +2 -0
              offset += tmp_step;
+
+             if (prev_done != done) {

Hmm... so this prints the status message every increase percentage point, right?

Not at all, the 'least change' it prints is one percent but that's the maximum granularity i.e. I tested with 10 GB file and the output was:

Discarding: 0% done
Discarding: 20% done
Discarding: 40% done
Discarding: 60% done
Discarding: 80% done
Discarding is done.

So ATM there could be up to 102 lines - please propose a different idea.

Quoted 5 lines

mkfs/xfs_mkfs.c discard_blocks(dev_t dev, uint64_t nsectors) · line 1276

Quoted 3 lines
mkfs/xfs_mkfs.c +2 -0
+                     prev_done = done;
+                     fprintf(stderr, _("Discarding: %2lu%% done\n"), done);

This isn't an error, so why output to stderr?

My bad, sorry.

Quoted 4 lines

FWIW if it's a tty you might consider ending that string with \r so the status messages don't scroll off the screen. Or possibly only reporting status if stdout is a tty?

Do I get it right that you propose to not flow the terminal with dozens of lines which just update the percentage but instead keep updating the same line? If so, I do like that.

Quoted 13 lines

--D

Quoted 9 lines
mkfs/xfs_mkfs.c +2 -1
+             }
      }
+     fprintf(stderr, _("Discarding is done.\n"));
 }
 
 static __attribute__((noreturn)) void
--

2.23.0

#17 lore raw

Re: [PATCH 2/2] mkfs: Show progress during block discard

Quoted 42 lines · Pavel Reichl
Quoted 30 lines · Darrick J. Wong
Quoted 28 lines · Pavel Reichl
---
 mkfs/xfs_mkfs.c | 8 ++++++++
 1 file changed, 8 insertions(+)
mkfs/xfs_mkfs.c +2 -0
index a02d6f66..07b8bd78 100644
@@ -1248,6 +1248,7 @@ discard_blocks(dev_t dev, uint64_t nsectors)
      const uint64_t  step            = (uint64_t)2<<30;
      /* Sector size is 512 bytes */
      const uint64_t  count           = nsectors << 9;
+     uint64_t        prev_done       = (uint64_t) ~0;
 
      fd = libxfs_device_to_fd(dev);
      if (fd <= 0)
@@ -1255,6 +1256,7 @@ discard_blocks(dev_t dev, uint64_t nsectors)
 
      while (offset < count) {
              uint64_t        tmp_step = step;
+             uint64_t        done = offset * 100 / count;
 
              if ((offset + step) > count)
                      tmp_step = count - offset;
@@ -1268,7 +1270,13 @@ discard_blocks(dev_t dev, uint64_t nsectors)
                      return;
 
Quoted 6 lines

mkfs/xfs_mkfs.c discard_blocks(dev_t dev, uint64_t nsectors) · line 1274

Quoted 3 lines
mkfs/xfs_mkfs.c +2 -0
              offset += tmp_step;
+
+             if (prev_done != done) {

Hmm... so this prints the status message every increase percentage point, right?

Not at all, the 'least change' it prints is one percent but that's the maximum granularity i.e. I tested with 10 GB file and the output was:

Quoted 8 lines

Discarding: 0% done
Discarding: 20% done
Discarding: 40% done
Discarding: 60% done
Discarding: 80% done
Discarding is done.

So ATM there could be up to 102 lines - please propose a different idea.

if (device supports discard) {
	if (!quiet)
		printf(_("Discarding blocks, this may take some time..."));
	<discard loop>
}

<the rest of mkfs>

Quoted 9 lines
Quoted 6 lines

mkfs/xfs_mkfs.c discard_blocks(dev_t dev, uint64_t nsectors) · line 1276

Quoted 4 lines
mkfs/xfs_mkfs.c +2 -0
+                     prev_done = done;
+                     fprintf(stderr, _("Discarding: %2lu%% done\n"), done);

This isn't an error, so why output to stderr?

My bad, sorry.

Quoted 1 line
Quoted 7 lines
Quoted 3 lines

FWIW if it's a tty you might consider ending that string with \r so the status messages don't scroll off the screen. Or possibly only reporting status if stdout is a tty?

Do I get it right that you propose to not flow the terminal with dozens of lines which just update the percentage but instead keep updating the same line? If so, I do like that.

Correct.

--D

Quoted 14 lines
Quoted 13 lines

--D

Quoted 9 lines
mkfs/xfs_mkfs.c +2 -1
+             }
      }
+     fprintf(stderr, _("Discarding is done.\n"));
 }
 
 static __attribute__((noreturn)) void
--

2.23.0

#18 lore raw

Re: [PATCH 2/2] mkfs: Show progress during block discard

In reply to Pavel Reichl
Quoted 18 lines · Pavel Reichl
---
 mkfs/xfs_mkfs.c | 8 ++++++++
 1 file changed, 8 insertions(+)
mkfs/xfs_mkfs.c +1 -0
index a02d6f66..07b8bd78 100644
@@ -1248,6 +1248,7 @@ discard_blocks(dev_t dev, uint64_t nsectors)
 	const uint64_t	step		= (uint64_t)2<<30;
 	/* Sector size is 512 bytes */
 	const uint64_t	count		= nsectors << 9;
+	uint64_t	prev_done	= (uint64_t) ~0;
 
 	fd = libxfs_device_to_fd(dev);
 	if (fd <= 0)

mkfs/xfs_mkfs.c discard_blocks(dev_t dev, uint64_t nsectors) · line 1259

Quoted 4 lines
mkfs/xfs_mkfs.c +1 -0
@@ -1255,6 +1256,7 @@ discard_blocks(dev_t dev, uint64_t nsectors)
 
 	while (offset < count) {
 		uint64_t	tmp_step = step;
+		uint64_t	done = offset * 100 / count;

That will overflow on a EB-scale (2^60 bytes) filesystems, won't it?

Quoted 4 lines
mkfs/xfs_mkfs.c +0 -0
 		if ((offset + step) > count)
 			tmp_step = count - offset;

mkfs/xfs_mkfs.c discard_blocks(dev_t dev, uint64_t nsectors) · line 1279

Quoted 10 lines
mkfs/xfs_mkfs.c +6 -0
@@ -1268,7 +1270,13 @@ discard_blocks(dev_t dev, uint64_t nsectors)
 			return;
 
 		offset += tmp_step;
+
+		if (prev_done != done) {
+			prev_done = done;
+			fprintf(stderr, _("Discarding: %2lu%% done\n"), done);
+		}
 	}
+	fprintf(stderr, _("Discarding is done.\n"));

Hmmm - this output doesn't get suppressed when the "quiet" (-q) option is used. mkfs is supposed to be silent when this option is specified.

I also suspect that it breaks a few fstests, too, as a some of them capture and filter mkfs output. They'll need filters to drop these new messages.

FWIW, a 100 lines of extra mkfs output is going to cause workflow issues. I know it will cause me problems, because I often mkfs 500TB filesystems tens of times a day on a discard enabled device. This extra output will scroll all the context of the previous test run I'm about to compare against off my terminal screen and so now I will have to scroll the terminal to look at the results of back-to-back runs. IOWs, I'm going to immediately want to turn this output off and have it stay off permanently.

Hence I think that, by default, just outputting a single "Discard in progress" line before starting the discard would be sufficient indication of what mkfs is currently doing. If someone wants more verbose progress output, then we should probably introduce a "verbose" CLI option to go along with the "quiet" option that suppresses all output.

Cheers,

Dave.

Signature · 2 lines
Dave Chinner
[email protected]
#19 lore raw

Re: [PATCH 2/2] mkfs: Show progress during block discard

Quoted 21 lines · Dave Chinner
Quoted 19 lines · Pavel Reichl
---
 mkfs/xfs_mkfs.c | 8 ++++++++
 1 file changed, 8 insertions(+)
mkfs/xfs_mkfs.c +1 -0
index a02d6f66..07b8bd78 100644
@@ -1248,6 +1248,7 @@ discard_blocks(dev_t dev, uint64_t nsectors)
      const uint64_t  step            = (uint64_t)2<<30;
      /* Sector size is 512 bytes */
      const uint64_t  count           = nsectors << 9;
+     uint64_t        prev_done       = (uint64_t) ~0;
 
      fd = libxfs_device_to_fd(dev);
      if (fd <= 0)
@@ -1255,6 +1256,7 @@ discard_blocks(dev_t dev, uint64_t nsectors)
 
Quoted 5 lines

mkfs/xfs_mkfs.c discard_blocks(dev_t dev, uint64_t nsectors) · line 1259

Quoted 3 lines
mkfs/xfs_mkfs.c +1 -0
      while (offset < count) {
              uint64_t        tmp_step = step;
+             uint64_t        done = offset * 100 / count;

That will overflow on a EB-scale (2^60 bytes) filesystems, won't it?

I guess that can happen, sorry. I'll try to come out with computation based on a floating point arithmetic. There should not be any performance or actual precision problem.
(well actually I'll drop this line completely, no ratio will be computed in the end)

Quoted 7 lines
Quoted 7 lines
mkfs/xfs_mkfs.c +0 -0
              if ((offset + step) > count)
                      tmp_step = count - offset;
@@ -1268,7 +1270,13 @@ discard_blocks(dev_t dev, uint64_t nsectors)
                      return;
 
Quoted 12 lines

mkfs/xfs_mkfs.c discard_blocks(dev_t dev, uint64_t nsectors) · line 1279

Quoted 8 lines
mkfs/xfs_mkfs.c +6 -0
              offset += tmp_step;
+
+             if (prev_done != done) {
+                     prev_done = done;
+                     fprintf(stderr, _("Discarding: %2lu%% done\n"), done);
+             }
      }
+     fprintf(stderr, _("Discarding is done.\n"));

Hmmm - this output doesn't get suppressed when the "quiet" (-q) option is used. mkfs is supposed to be silent when this option is specified.

OK, my bad. I'll fix that.

Quoted 14 lines

I also suspect that it breaks a few fstests, too, as a some of them capture and filter mkfs output. They'll need filters to drop these new messages.

FWIW, a 100 lines of extra mkfs output is going to cause workflow issues. I know it will cause me problems, because I often mkfs 500TB filesystems tens of times a day on a discard enabled device. This extra output will scroll all the context of the previous test run I'm about to compare against off my terminal screen and so now I will have to scroll the terminal to look at the results of back-to-back runs. IOWs, I'm going to immediately want to turn this output off and have it stay off permanently.

Quoted 2 lines

Hence I think that, by default, just outputting a single "Discard in progress" line before starting the discard would be sufficient

OK, maybe just one line "Discard in progress" is actually what users need. The computing of % done was probably just overkill from my side. Sorry about that.

Quoted 12 lines

indication of what mkfs is currently doing. If someone wants more verbose progress output, then we should probably introduce a "verbose" CLI option to go along with the "quiet" option that suppresses all output.

Cheers,

Dave.
--
Dave Chinner
[email protected]

#20 lore raw

Re: [PATCH 2/2] mkfs: Show progress during block discard

Quoted 28 lines · Pavel Reichl
Quoted 21 lines · Dave Chinner
Quoted 19 lines · Pavel Reichl
---
 mkfs/xfs_mkfs.c | 8 ++++++++
 1 file changed, 8 insertions(+)
mkfs/xfs_mkfs.c +1 -0
index a02d6f66..07b8bd78 100644
@@ -1248,6 +1248,7 @@ discard_blocks(dev_t dev, uint64_t nsectors)
      const uint64_t  step            = (uint64_t)2<<30;
      /* Sector size is 512 bytes */
      const uint64_t  count           = nsectors << 9;
+     uint64_t        prev_done       = (uint64_t) ~0;
 
      fd = libxfs_device_to_fd(dev);
      if (fd <= 0)
@@ -1255,6 +1256,7 @@ discard_blocks(dev_t dev, uint64_t nsectors)
 
Quoted 5 lines

mkfs/xfs_mkfs.c discard_blocks(dev_t dev, uint64_t nsectors) · line 1259

Quoted 3 lines
mkfs/xfs_mkfs.c +1 -0
      while (offset < count) {
              uint64_t        tmp_step = step;
+             uint64_t        done = offset * 100 / count;

That will overflow on a EB-scale (2^60 bytes) filesystems, won't it?

Quoted 5 lines

I guess that can happen, sorry. I'll try to come out with computation based on a floating point arithmetic. There should not be any performance or actual precision problem.
(well actually I'll drop this line completely, no ratio will be computed in the end)

No need to apologise for not realising huge filesystems need to work. It takes time to get used to having to consider 64 bit overflows everywhere... :)

Maybe the easiest way to do this sort of thing is to calculate reporting interval prior to the loop, and every time it is exceeded issue a report and then reset the report counter to zero. No fancy math required there. If we want 1% increments:

	report_interval = count / 100;
	
	while (offset < count) {
	....
		offset += tmp_step;
		report_offset += tmp_step;

		if (report_offset > report_interval) {
			report_offset = 0;
			/* issue report */
		}
	}

And this is easy to adjust the number of reports issued (e.g. every 5% or 10% is just changing the report_interval division constant.

Another way of doing it is deciding on the -time- between reports. e.g. issue a progress report every 60s. Then you can just report the percentage done based on offset and count without needing intermediate accounting.

Quoted 13 lines
Quoted 13 lines

I also suspect that it breaks a few fstests, too, as a some of them capture and filter mkfs output. They'll need filters to drop these new messages.

FWIW, a 100 lines of extra mkfs output is going to cause workflow issues. I know it will cause me problems, because I often mkfs 500TB filesystems tens of times a day on a discard enabled device. This extra output will scroll all the context of the previous test run I'm about to compare against off my terminal screen and so now I will have to scroll the terminal to look at the results of back-to-back runs. IOWs, I'm going to immediately want to turn this output off and have it stay off permanently.

Quoted 6 lines
Quoted 2 lines

Hence I think that, by default, just outputting a single "Discard in progress" line before starting the discard would be sufficient

OK, maybe just one line "Discard in progress" is actually what users need. The computing of % done was probably just overkill from my side. Sorry about that.

Again, no need to apologise because there are different opinions on how something should be done. If you didn't put progress reporting in, I'm sure someone would have suggested it and we'd be having the same discussion anyway. :)

Cheers,

Dave.

Signature · 2 lines
Dave Chinner
[email protected]
#21 lore raw

Re: [PATCH 2/2] mkfs: Show progress during block discard

In reply to Pavel Reichl
Quoted 3 lines · Pavel Reichl
Quoted 2 lines · Dave Chinner

On Thu, Nov 21, 2019 at 10:44:45PM +0100, Pavel Reichl wrote:

...

Quoted 14 lines
Quoted 14 lines

I also suspect that it breaks a few fstests, too, as a some of them capture and filter mkfs output. They'll need filters to drop these new messages.

FWIW, a 100 lines of extra mkfs output is going to cause workflow issues. I know it will cause me problems, because I often mkfs 500TB filesystems tens of times a day on a discard enabled device. This extra output will scroll all the context of the previous test run I'm about to compare against off my terminal screen and so now I will have to scroll the terminal to look at the results of back-to-back runs. IOWs, I'm going to immediately want to turn this output off and have it stay off permanently.

Quoted 2 lines
Quoted 2 lines

Hence I think that, by default, just outputting a single "Discard in progress" line before starting the discard would be sufficient

e2fsprogs simply does:

Discarding device blocks: done

("done" isn't printed until it's ... done)

so that might be a good convention to follow? Though I'd probably do

printf("Discarding blocks... ");
....
printf("Done.\n");

because the ellipses tend to indicate waiting. :)

Even the one line might require filtering-out in xfstests, but luckily we have standard filters and it should be trivial to add.

Quoted 3 lines

OK, maybe just one line "Discard in progress" is actually what users need. The computing of % done was probably just overkill from my side. Sorry about that.

No worries, that's why we discuss stuff. :)
Thanks for taking this on,

-Eric