vhost/scsi: use vmalloc for order-10 allocation
Subsystems (6)
Michael S. Tsirkin
maintainerPATCH 1/1 vhost/scsi: use vmalloc for order-10 allocation
As vhost scsi device struct is large, if the device is created on a busy system, kzalloc() might fail, so this patch does a fallback to vzalloc().
As vmalloc() adds overhead on data-path, add __GFP_REPEAT to kzalloc() flags to do this fallback only when really needed.
---
I put this on my vhost fixes branch, intend to merge for 3.12. Dan, could you please confirm this works for you?
drivers/vhost/scsi.c | 41 +++++++++++++++++++++++++++-------------- 1 file changed, 27 insertions(+), 14 deletions(-)
drivers/vhost/scsi.c +27 -14
@@ -1373,21 +1373,30 @@ static int vhost_scsi_set_features(struct vhost_scsi *vs, u64 features)return 0 ; } +static void vhost_scsi_free (struct vhost_scsi *vs) +{ +if (is_vmalloc_addr (vs)) +vfree (vs); +else +kfree (vs); +} +static int vhost_scsi_open (struct inode *inode,struct file *f) {struct vhost_scsi *vs;struct vhost_virtqueue **vqs; -int r, i; +int r = -ENOMEM , i; - vs =kzalloc (sizeof (*vs),GFP_KERNEL ); -if (!vs) -return -ENOMEM ; + vs =kzalloc (sizeof (*vs),GFP_KERNEL | __GFP_NOWARN | __GFP_REPEAT); +if (!vs) { + vs =vzalloc (sizeof (*vs)); +if (!vs) + goto err_vs; + } vqs =kmalloc (VHOST_SCSI_MAX_VQ *sizeof (*vqs),GFP_KERNEL ); -if (!vqs) { -kfree (vs); -return -ENOMEM ; - } +if (!vqs) + goto err_vqs;vhost_work_init (&vs->vs_completion_work, vhost_scsi_complete_cmd_work);vhost_work_init (&vs->vs_event_work, tcm_vhost_evt_work);
@@ -1407,14 +1416,18 @@ static int vhost_scsi_open(struct inode *inode, struct file *f)tcm_vhost_init_inflight (vs,NULL ); -if (r <0 ) { -kfree (vqs); -kfree (vs); -return r; - } +if (r <0 ) + goto err_init; f->private_data = vs;return 0 ; + +err_init: +kfree (vqs); +err_vqs: +vhost_scsi_free (vs); +err_vs: +return r; }static int vhost_scsi_release (struct inode *inode,struct file *f)
@@ -1431,7 +1444,7 @@ static int vhost_scsi_release(struct inode *inode, struct file *f)/* Jobs can re-queue themselves in evt kick handler. Do extra flush. */ vhost_scsi_flush (vs);kfree (vs->dev.vqs); -kfree (vs); +vhost_scsi_free (vs);return 0 ; }
Signature · 1 line
MST
Quoted 9 lines · Michael S. Tsirkin
As vhost scsi device struct is large, if the device is created on a busy system,
kzalloc()might fail, so this patch does a fallback tovzalloc().As
vmalloc()adds overhead on data-path, add__GFP_REPEATtokzalloc()flags to do this fallback only when really needed.Reported-by Dan Aloni <[email protected]>Signed-off-by Michael S. Tsirkin <[email protected]>
Quoted 86 lines
---
I put this on my vhost fixes branch, intend to merge for 3.12. Dan, could you please confirm this works for you?
drivers/vhost/scsi.c | 41 +++++++++++++++++++++++++++-------------- 1 file changed, 27 insertions(+), 14 deletions(-)drivers/vhost/scsi.c +27
-14@@ -1373,21 +1373,30 @@ static int vhost_scsi_set_features(struct vhost_scsi *vs, u64 features)return 0 ; } +static void vhost_scsi_free (struct vhost_scsi *vs) +{ +if (is_vmalloc_addr (vs)) +vfree (vs); +else +kfree (vs); +} +static int vhost_scsi_open (struct inode *inode,struct file *f) {struct vhost_scsi *vs;struct vhost_virtqueue **vqs; -int r, i; +int r = -ENOMEM , i; - vs =kzalloc (sizeof (*vs),GFP_KERNEL ); -if (!vs) -return -ENOMEM ; + vs =kzalloc (sizeof (*vs),GFP_KERNEL | __GFP_NOWARN | __GFP_REPEAT); +if (!vs) { + vs =vzalloc (sizeof (*vs)); +if (!vs) + goto err_vs; + } vqs =kmalloc (VHOST_SCSI_MAX_VQ *sizeof (*vqs),GFP_KERNEL ); -if (!vqs) { -kfree (vs); -return -ENOMEM ; - } +if (!vqs) + goto err_vqs;vhost_work_init (&vs->vs_completion_work, vhost_scsi_complete_cmd_work);vhost_work_init (&vs->vs_event_work, tcm_vhost_evt_work);@@ -1407,14 +1416,18 @@ static int vhost_scsi_open(struct inode *inode, struct file *f)tcm_vhost_init_inflight (vs,NULL ); -if (r <0 ) { -kfree (vqs); -kfree (vs); -return r; - } +if (r <0 ) + goto err_init; f->private_data = vs;return 0 ; + +err_init: +kfree (vqs); +err_vqs: +vhost_scsi_free (vs); +err_vs: +return r; }static int vhost_scsi_release (struct inode *inode,struct file *f)@@ -1431,7 +1444,7 @@ static int vhost_scsi_release(struct inode *inode, struct file *f)/* Jobs can re-queue themselves in evt kick handler. Do extra flush. */ vhost_scsi_flush (vs);kfree (vs->dev.vqs); -kfree (vs); +vhost_scsi_free (vs);return 0 ; }Signature · 1 line
MST
Signature · 1 line
Asias
Hello.
Quoted 22 lines · Michael S. Tsirkin
As vhost scsi device struct is large, if the device is created on a busy system,
kzalloc()might fail, so this patch does a fallback tovzalloc().As
vmalloc()adds overhead on data-path, add__GFP_REPEATtokzalloc()flags to do this fallback only when really needed.Reported-by Dan Aloni <[email protected]>Signed-off-by Michael S. Tsirkin <[email protected]>---
I put this on my vhost fixes branch, intend to merge for 3.12. Dan, could you please confirm this works for you?
drivers/vhost/scsi.c | 41 +++++++++++++++++++++++++++-------------- 1 file changed, 27 insertions(+), 14 deletions(-)drivers/vhost/scsi.c +0
-0
drivers/vhost/scsi.c static int vhost_scsi_set_features(struct vhost_scsi *vs, u64 features) · line 1381
Quoted 9 lines
drivers/vhost/scsi.c +6
-0@@ -1373,21 +1373,30 @@ static int vhost_scsi_set_features(struct vhost_scsi *vs, u64 features)return 0 ; } +static void vhost_scsi_free (struct vhost_scsi *vs) +{ +if (is_vmalloc_addr (vs)) +vfree (vs); +else +kfree (vs);
Indent with the tabs ISO spaces, please.
drivers/vhost/scsi.c static int vhost_scsi_set_features(struct vhost_scsi *vs, u64 features) · line 1390
Quoted 13 lines
drivers/vhost/scsi.c +4
-4+} +static int vhost_scsi_open (struct inode *inode,struct file *f) {struct vhost_scsi *vs;struct vhost_virtqueue **vqs; -int r, i; +int r = -ENOMEM , i; - vs =kzalloc (sizeof (*vs),GFP_KERNEL ); -if (!vs) -return -ENOMEM ; + vs =kzalloc (sizeof (*vs),GFP_KERNEL | __GFP_NOWARN | __GFP_REPEAT);
Indent here with a tab, please.
drivers/vhost/scsi.c static int vhost_scsi_set_features(struct vhost_scsi *vs, u64 features) · line 1395
Quoted 5 lines
drivers/vhost/scsi.c +5
-0+if (!vs) { + vs =vzalloc (sizeof (*vs)); +if (!vs) + goto err_vs; + }
WBR, Sergei
Michael S. Tsirkin
maintainerQuoted 54 lines · Sergei Shtylyov
Hello.
Quoted 22 lines · Michael S. Tsirkin
As vhost scsi device struct is large, if the device is created on a busy system,
kzalloc()might fail, so this patch does a fallback tovzalloc().As
vmalloc()adds overhead on data-path, add__GFP_REPEATtokzalloc()flags to do this fallback only when really needed.Reported-by Dan Aloni <[email protected]>Signed-off-by Michael S. Tsirkin <[email protected]>---
I put this on my vhost fixes branch, intend to merge for 3.12. Dan, could you please confirm this works for you?
drivers/vhost/scsi.c | 41 +++++++++++++++++++++++++++-------------- 1 file changed, 27 insertions(+), 14 deletions(-)drivers/vhost/scsi.c +0
-0drivers/vhost/scsi.c static int vhost_scsi_set_features(struct vhost_scsi *vs, u64 features) · line 1381
Quoted 9 lines
drivers/vhost/scsi.c +6
-0@@ -1373,21 +1373,30 @@ static int vhost_scsi_set_features(struct vhost_scsi *vs, u64 features)return 0 ; } +static void vhost_scsi_free (struct vhost_scsi *vs) +{ +if (is_vmalloc_addr (vs)) +vfree (vs); +else +kfree (vs);Indent with the tabs ISO spaces, please.
drivers/vhost/scsi.c static int vhost_scsi_set_features(struct vhost_scsi *vs, u64 features) · line 1390
Quoted 13 lines
drivers/vhost/scsi.c +4
-4+} +static int vhost_scsi_open (struct inode *inode,struct file *f) {struct vhost_scsi *vs;struct vhost_virtqueue **vqs; -int r, i; +int r = -ENOMEM , i; - vs =kzalloc (sizeof (*vs),GFP_KERNEL ); -if (!vs) -return -ENOMEM ; + vs =kzalloc (sizeof (*vs),GFP_KERNEL | __GFP_NOWARN | __GFP_REPEAT);Indent here with a tab, please.
Quoted 7 lines
drivers/vhost/scsi.c static int vhost_scsi_set_features(struct vhost_scsi *vs, u64 features) · line 1395
Quoted 5 lines
drivers/vhost/scsi.c +5
-0+if (!vs) { + vs =vzalloc (sizeof (*vs)); +if (!vs) + goto err_vs; + }WBR, Sergei
Thanks, I'll fix this up.