Adds persistent entryies using request_firmware_nowaitManuel Estrada Sainz <[email protected]>,
Subsystems (2)
PATCH 1/1 Adds persistent entryies using request_firmware_nowaitManuel Estrada Sainz <[email protected]>,
This is a patch to make the /sys/class/firmware entries persistent. This has been tested with dell_rbu; dell_rbu was modified to not call request_firmware_nowait again form the callback function.
The new mechanism to make the entries persistent is as follows 1> echo 0 > /sys/class/firmware/timeout
2> echo 2 > /sys/class/firmware/xxx/loading
step 1 prevents timeout to occur , step 2 makes the entry xxx persistent
if we want to remove persistence then do this
ech0 -2 > /sys/class/firmware/xxx/loading
The rest of the functionality is not affected.
Also not the persistence is supported only if the driver calls request_firmware_nowait. If the driver is just calling request_firmware, step 2 is treated as unknown entry.
Thanks,
Abhay Salunke
Software Engineer.
DELL Inc
usr/src/linux-2.6.11.11.new/drivers/base/firmware_class.c renamed from usr/src/linux-2.6.11.11.orig/drivers/base/firmware_class.c +53 -6
@@ -6,6 +6,11 @@* Please see Documentation/firmware_class/ for more information. * */ +/* +* 2005-06-15: Abhay Salunke <[email protected]> +* Added firmware persistent when request_firmware_nowait. +* is called. +*/ #include <linux/device.h> #include <linux/module.h>
@@ -28,6 +33,8 @@ enum {FW_STATUS_DONE ,FW_STATUS_ABORT ,FW_STATUS_READY , +FW_STATUS_PERSISTENT , + FW_STATUS_PERSISTENT_ABORT, };static int loading_timeout =10 ;/* In seconds */
@@ -142,13 +149,20 @@ firmware_loading_store(struct class_deviset_bit (FW_STATUS_LOADING , &fw_priv->status);up (&fw_lock);break ; +case 2 : +set_bit (FW_STATUS_PERSISTENT , &fw_priv->status); +fw_load_abort (fw_priv); +break ;case 0 :if (test_bit (FW_STATUS_LOADING , &fw_priv->status)) {complete (&fw_priv->completion);clear_bit (FW_STATUS_LOADING , &fw_priv->status);break ; } -/* fallthrough */ +case -2 : +set_bit (FW_STATUS_PERSISTENT_ABORT , &fw_priv->status); +fw_load_abort (fw_priv); +break ;default :printk (KERN_ERR "%s: unexpected value (%d)\n" , __FUNCTION__, loading);
@@ -389,8 +403,8 @@ out:* firmware image for this or any other device. **/ int -request_firmware (const struct firmware **firmware_p,const char *name, -struct device *device) +_request_firmware (const struct firmware **firmware_p,const char *name, +struct device *device,unsigned long *fw_status) {struct class_device *class_dev;struct firmware_priv *fw_priv;
@@ -422,6 +436,14 @@ request_firmware(const struct firmware *kobject_hotplug (&class_dev->kobj,KOBJ_ADD );wait_for_completion (&fw_priv->completion); + +if (test_bit (FW_STATUS_PERSISTENT , &fw_priv->status) || +test_bit (FW_STATUS_PERSISTENT_ABORT , &fw_priv->status)) { + *fw_status = fw_priv->status; +clear_bit (FW_STATUS_PERSISTENT , &fw_priv->status); +clear_bit (FW_STATUS_PERSISTENT_ABORT , &fw_priv->status); + } +set_bit (FW_STATUS_DONE , &fw_priv->status);del_timer_sync (&fw_priv->timeout);
@@ -443,6 +465,25 @@ error_kfree_fw: out:return retval; } +/** +* request_firmware: - request firmware to hotplug and wait for it +* Description: +* @firmware will be used to return a firmware image by the name +* of @name for device @device. +* +* Should be called from user context where sleeping is allowed. +* +* @name will be use as $FIRMWARE in the hotplug environment and +* should be distinctive enough not to be confused with any other +* firmware image for this or any other device. +**/ +int +request_firmware (const struct firmware **firmware_p,const char *name, +struct device *device) +{ +unsigned long status; +return _request_firmware (firmware_p, name,device,&status); +}/** * release_firmware: - release the resource associated with a firmware image
@@ -481,6 +522,7 @@ struct firmware_work {struct device *device;void *context;void (*cont)(const struct firmware *fw,void *context); +unsigned long status; }; static int
@@ -493,9 +535,14 @@ request_firmware_work_func(void *arg)return 0 ; }daemonize ("%s/%s" ,"firmware" , fw_work->name); -request_firmware (&fw, fw_work->name, fw_work->device); - fw_work->cont (fw, fw_work->context); -release_firmware (fw); + fw_work->status =FW_STATUS_LOADING ; + +do { +_request_firmware (&fw, fw_work->name, fw_work->device, &fw_work->status); + fw_work->cont (fw, fw_work->context); +release_firmware (fw); + }while (test_bit (FW_STATUS_PERSISTENT , &fw_work->status)); +module_put (fw_work->module);kfree (fw_work);return 0 ;
Quoted 3 lines · Abhay Salunke
---
/usr/src/linux-2.6.11.11.orig/drivers/base/firmware_class.c2005-06-14 20:53:04.000000000 -0500
+++/usr/src/linux-2.6.11.11.new/drivers/base/firmware_class.c2005-06-16 00:21:10.000000000 -0500
- Please prepare patches in `patch -p1' form, as per
http://www.zip.com.au/~akpm/linux/patches/stuff/tpp.txt
- Please use hard tabs, not a mixture of spaces and hard tabs
- Try to avoid adding new trailing whitespace.
Here's a cleaned-up patch:
From: Abhay Salunke <[email protected]>
This is a patch to make the /sys/class/firmware entries persistent. This has been tested with dell_rbu; dell_rbu was modified to not call request_firmware_nowait again form the callback function.
The new mechanism to make the entries persistent is as follows 1> echo 0 > /sys/class/firmware/timeout
2> echo 2 > /sys/class/firmware/xxx/loading
step 1 prevents timeout to occur , step 2 makes the entry xxx persistent
if we want to remove persistence then do this
ech0 -2 > /sys/class/firmware/xxx/loading
The rest of the functionality is not affected.
Also not the persistence is supported only if the driver calls request_firmware_nowait. If the driver is just calling request_firmware, step 2 is treated as unknown entry.
---
drivers/base/firmware_class.c | 59 +++++++++++++++++++++++++++++++++++++----- 1 files changed, 53 insertions(+), 6 deletions(-)
drivers/base/firmware_class.c renamed from drivers/base/firmware_class.c~firmware-add-persistent-entries-using-request_firmware_nowait +53 -6
@@ -6,6 +6,11 @@* Please see Documentation/firmware_class/ for more information. * */ +/* +* 2005-06-15: Abhay Salunke <[email protected]> +* Added firmware persistent when request_firmware_nowait. +* is called. +*/ #include <linux/device.h> #include <linux/module.h>
@@ -28,6 +33,8 @@ enum {FW_STATUS_DONE ,FW_STATUS_ABORT ,FW_STATUS_READY , +FW_STATUS_PERSISTENT , + FW_STATUS_PERSISTENT_ABORT, };static int loading_timeout =10 ;/* In seconds */
@@ -145,13 +152,20 @@ firmware_loading_store(struct class_deviset_bit (FW_STATUS_LOADING , &fw_priv->status);up (&fw_lock);break ; +case 2 : +set_bit (FW_STATUS_PERSISTENT , &fw_priv->status); +fw_load_abort (fw_priv); +break ;case 0 :if (test_bit (FW_STATUS_LOADING , &fw_priv->status)) {complete (&fw_priv->completion);clear_bit (FW_STATUS_LOADING , &fw_priv->status);break ; } -/* fallthrough */ +case -2 : +set_bit (FW_STATUS_PERSISTENT_ABORT , &fw_priv->status); +fw_load_abort (fw_priv); +break ;default :printk (KERN_ERR "%s: unexpected value (%d)\n" , __FUNCTION__, loading);
@@ -392,8 +406,8 @@ out:* firmware image for this or any other device. **/ int -request_firmware (const struct firmware **firmware_p,const char *name, -struct device *device) +_request_firmware (const struct firmware **firmware_p,const char *name, +struct device *device,unsigned long *fw_status) {struct class_device *class_dev;struct firmware_priv *fw_priv;
@@ -425,6 +439,14 @@ request_firmware(const struct firmware *kobject_hotplug (&class_dev->kobj,KOBJ_ADD );wait_for_completion (&fw_priv->completion); + +if (test_bit (FW_STATUS_PERSISTENT , &fw_priv->status) || +test_bit (FW_STATUS_PERSISTENT_ABORT , &fw_priv->status)) { + *fw_status = fw_priv->status; +clear_bit (FW_STATUS_PERSISTENT , &fw_priv->status); +clear_bit (FW_STATUS_PERSISTENT_ABORT , &fw_priv->status); + } +set_bit (FW_STATUS_DONE , &fw_priv->status);del_timer_sync (&fw_priv->timeout);
@@ -446,6 +468,25 @@ error_kfree_fw: out:return retval; } +/** +* request_firmware: - request firmware to hotplug and wait for it +* Description: +* @firmware will be used to return a firmware image by the name +* of @name for device @device. +* +* Should be called from user context where sleeping is allowed. +* +* @name will be use as $FIRMWARE in the hotplug environment and +* should be distinctive enough not to be confused with any other +* firmware image for this or any other device. +**/ +int +request_firmware (const struct firmware **firmware_p,const char *name, +struct device *device) +{ +unsigned long status; +return _request_firmware (firmware_p, name,device,&status); +}/** * release_firmware: - release the resource associated with a firmware image
@@ -484,6 +525,7 @@ struct firmware_work {struct device *device;void *context;void (*cont)(const struct firmware *fw,void *context); +unsigned long status; }; static int
@@ -496,9 +538,14 @@ request_firmware_work_func(void *arg)return 0 ; }daemonize ("%s/%s" ,"firmware" , fw_work->name); -request_firmware (&fw, fw_work->name, fw_work->device); - fw_work->cont (fw, fw_work->context); -release_firmware (fw); + fw_work->status =FW_STATUS_LOADING ; + +do { +_request_firmware (&fw, fw_work->name, fw_work->device, &fw_work->status); + fw_work->cont (fw, fw_work->context); +release_firmware (fw); + }while (test_bit (FW_STATUS_PERSISTENT , &fw_work->status)); +module_put (fw_work->module);kfree (fw_work);return 0 ;
_
Quoted 4 lines · Abhay Salunke
This is a patch to make the
/sys/class/firmwareentries persistent. This has been tested withdell_rbu;dell_rbuwas modified to not callrequest_firmware_nowaitagain form the callback function.
Quoted 9 lines
The new mechanism to make the entries persistent is as follows 1> echo 0 >
/sys/class/firmware/timeout
2> echo 2 >/sys/class/firmware/xxx/loadingstep 1 prevents timeout to occur , step 2 makes the entry xxx persistent
if we want to remove persistence then do this
ech0 -2 >/sys/class/firmware/xxx/loading
Hi,
I have the following issues with the patch:
- since "persistency" (or rather repeat loading) is controlled from userspace, drivers don't have control over it. This way every user of request_firmware_nowait has to be ready to process more than one firmware load.
- There is no way to "cancel" firmware request from the driver. You will not be able to safely unload users of request_firmware_nowait(). Since loader is rearming you can't use firmware handler function to signal when request has been processed.
I think that such re-arming reqests are much better implemented in individual drivers.
Signature · 1 line
Dmitry
Re: [patch 2.6.12-rc3] Adds persistent entryies using request_firmware_nowait
Quoted 15 lines · Dmitry Torokhov
Quoted 4 lines · Abhay Salunke
This is a patch to make the
/sys/class/firmwareentries persistent. This has been tested withdell_rbu;dell_rbuwas modified to not callrequest_firmware_nowaitagain form the callback function.Quoted 9 lines
The new mechanism to make the entries persistent is as follows 1> echo 0 >
/sys/class/firmware/timeout
2> echo 2 >/sys/class/firmware/xxx/loadingstep 1 prevents timeout to occur , step 2 makes the entry xxx persistent
if we want to remove persistence then do this
ech0 -2 >/sys/class/firmware/xxx/loading
Quoted 8 lines
Hi,
I have the following issues with the patch:
- since "persistency" (or rather repeat loading) is controlled from userspace, drivers don't have control over it. This way every user of
request_firmware_nowaithas to be ready to process more than one firmware load.
The user has knowingly choosen to be persistent and it will be responsible for handling multiple requests. I understand the concern here is that if by accident the user writes 2 to loading...I am working on the next patch which will address this issue.
Quoted 6 lines
- There is no way to "cancel" firmware request from the driver. You will not be able to safely unload users of
request_firmware_nowait(). Since loader is rearming you can't use firmware handler function to signal when request has been processed.
This is a valid concern and it is also true with the current code. Calling request_firmware_nowait for the current code the driver is at mercy of the user to send a cancel request to loading. Any driver unload will fail till the user cancles it.
I realized that while playing with dell_rbu ( which uses request_firmware_nowait)
Quoted 2 lines
I think that such re-arming reqests are much better implemented in individual drivers.
I respecfully disagree ; I think the request_firmware_nowait is not complete if we dont have a way to make it persistent. Also if the other drivers are required to do the re-arming they will still end up in the same situation of not being able to unload unless the user chooses to cancel firmware. The best fix is to fix this in frimware_class.c.
I had experienced this exact thing with dell_rbu driver.I will address these issues in my next patch.
Thanks
Abhay
Quoted 4 lines · Abhay Salunke
This is a patch to make the
/sys/class/firmwareentries persistent. This has been tested withdell_rbu;dell_rbuwas modified to not callrequest_firmware_nowaitagain form the callback function.
Quoted 8 lines
The new mechanism to make the entries persistent is as follows 1> echo 0 >
/sys/class/firmware/timeout
2> echo 2 >/sys/class/firmware/xxx/loadingstep 1 prevents timeout to occur , step 2 makes the entry xxx persistent
if we want to remove persistence then do this
ech0 -2 >/sys/class/firmware/xxx/loading
Hm, those are some mighty "magic" numbers that will be tough for people to realize exactly what they mean. Try adding a "persistant" file instead.
Quoted 5 lines
+ /*
+ * 2005-06-15: Abhay Salunke <[email protected]>
+ * Added firmware persistent whenrequest_firmware_nowait.
+ * is called.
+ */
Don't add changelog comments to .c files. That belongs in the git tree, not in the code itself.
Also, your use of tabs and spaces are wrong in a lot of places...
thanks,
greg k-h