x86 BIOS interface for partitioning and system serial number on SGI UV
x86 BIOS interface for partitioning and system serial number on SGI UV
--- arch/x86/kernel/Makefile | 2 arch/x86/kernel/bios_uv.c | 87 +++++++++++++++++++++++++- arch/x86/kernel/genx2apic_uv_x.c | 8 +- arch/x86/kernel/tlb_uv.c | 5 + arch/x86/kernel/uv_proc_fs.c | 122 +++++++++++++++++++++++++++++++++++++ include/asm-x86/uv/bios.h | 127 ++++++++++++++++++++++++++++++++++----- 6 files changed, 326 insertions(+), 25 deletions(-)
arch/x86/kernel/tlb_uv.c +4 -1
@@ -14,6 +14,7 @@#include <asm/uv/uv_mmrs.h> #include <asm/uv/uv_hub.h> #include <asm/uv/uv_bau.h> +#include <asm/uv/bios.h> #include <asm/genapic.h> #include <asm/idle.h> #include <asm/tsc.h>
@@ -565,7 +566,9 @@ static int __init uv_ptc_init(void)if (!is_uv_system ())return 0 ; -if (!proc_mkdir ("sgi_uv" ,NULL )) +if (!sgi_uv_dir) + sgi_uv_dir =proc_mkdir ("sgi_uv" ,NULL ); +if (!sgi_uv_dir)return -EINVAL ; proc_uv_ptc =create_proc_entry (UV_PTC_BASENAME ,0444 ,NULL );
arch/x86/kernel/uv_proc_fs.c new file +122 -0
122 lines not on this page · raw message
arch/x86/kernel/Makefile +1 -1
@@ -104,7 +104,7 @@ obj-$(CONFIG_OLPC) += olpc.o # 64 bit specific files ifeq ($(CONFIG_X86_64),y) obj-y += genapic_64.o genapic_flat_64.o genx2apic_uv_x.o tlb_uv.o - obj-y += bios_uv.o + obj-y += bios_uv.o uv_proc_fs.o obj-y += genx2apic_cluster.o obj-y += genx2apic_phys.o obj-$(CONFIG_X86_PM_TIMER) += pmtimer_64.o
arch/x86/kernel/bios_uv.c +82 -5
113 lines not on this page · raw message
arch/x86/kernel/genx2apic_uv_x.c +6 -2
@@ -344,7 +344,7 @@ static __init void uv_rtc_init(void)long status, ticks_per_sec, drift; status = -x86_bios_freq_base (BIOS_FREQ_BASE_REALTIME_CLOCK , &ticks_per_sec, +uv_bios_freq_base (BIOS_FREQ_BASE_REALTIME_CLOCK , &ticks_per_sec, &drift);if (status !=0 || ticks_per_sec <100000 ) { printk(KERN_WARNING
@@ -410,6 +410,10 @@ static __init void uv_system_init(void) gnode_upper = (((unsigned long )node_id.s.node_id) & ~((1 << n_val) -1 )) << m_val; +uv_bios_get_sn_info (0 , &uv_type, &uv_nasid_bitmask, + &uv_nasid_shift, &uv_system_size, + &uv_sharing_domain_size, &sn_partition_id, + &uv_coherency_id, &uv_region_size);uv_rtc_init ();for_each_present_cpu (cpu) {
@@ -431,7 +435,7 @@ static __init void uv_system_init(void)uv_cpu_hub_info (cpu)->gpa_mask = (1 << (m_val + n_val)) -1 ;uv_cpu_hub_info (cpu)->gnode_upper = gnode_upper;uv_cpu_hub_info (cpu)->global_mmr_base = mmr_base; -uv_cpu_hub_info (cpu)->coherency_domain_number =0 ;/* ZZZ */ +uv_cpu_hub_info (cpu)->coherency_domain_number = uv_coherency_id; uv_node_to_blade[nid] = blade; uv_cpu_to_blade[cpu] = blade; max_pnode =max (pnode, max_pnode);
include/asm-x86/uv/bios.h +111 -16
156 lines not on this page · raw message
Signature · 2 lines
Russ Anderson, OS RAS/Partitioning Project Lead SGI - Silicon Graphics Inc [email protected]
H. Peter Anvin
reviewerarch/x86/kernel/tlb_uv.c static int __init uv_ptc_init(void) · line 572
Quoted 7 lines · Russ Anderson
x86 BIOS interface for partitioning and system serial number on SGI UV
arch/x86/kernel/tlb_uv.c +3
-1-if (!proc_mkdir ("sgi_uv" ,NULL )) +if (!sgi_uv_dir) + sgi_uv_dir =proc_mkdir ("sgi_uv" ,NULL ); +if (!sgi_uv_dir)return -EINVAL ;
This really seems more that it should belong in sysfs -- /sys/class/firmware presumably.
include/asm-x86/uv/bios.h struct uv_bios_retval { · line 66
Quoted 7 lines
include/asm-x86/uv/bios.h +7
-0+# define BIOS_CALL (result, a0, a1, a2, a3, a4, a5, a6, a7) \ + do { \ +/* XXX - the real call goes here */ \ + result.status =BIOS_STATUS_UNIMPLEMENTED ; \ + result.v0 =0 ; \ + result.v1 =0 ; \ + }while (0 )
I have more than a little problem with submitting patches like this. We have no way to judge the suitability of the coding or the interface with the "meat" of the driver stubbed out!
-hpa
Quoted 11 lines · H. Peter Anvin
Quoted 7 lines · Russ Anderson
x86 BIOS interface for partitioning and system serial number on SGI UV
arch/x86/kernel/tlb_uv.c +3
-1-if (!proc_mkdir ("sgi_uv" ,NULL )) +if (!sgi_uv_dir) + sgi_uv_dir =proc_mkdir ("sgi_uv" ,NULL ); +if (!sgi_uv_dir)return -EINVAL;This really seems more that it should belong in sysfs --
/sys/class/firmwarepresumably.
/proc/sgi_uv already exists, similar to /proc/sgi_sn on Itanium systems.
Would it be /sys/class/firmware? Reading Documentation/sysfs-rules.txt seems to indicate somewhere under /sys/devices, such as /sys/devices/system/, but I may be wrong.
Quoted 11 lines
include/asm-x86/uv/bios.h struct uv_bios_retval { · line 66
Quoted 7 lines
include/asm-x86/uv/bios.h +7
-0+# define BIOS_CALL (result, a0, a1, a2, a3, a4, a5, a6, a7) \ + do { \ +/* XXX - the real call goes here */ \ + result.status =BIOS_STATUS_UNIMPLEMENTED ; \ + result.v0 =0 ; \ + result.v1 =0 ; \ + }while (0 )I have more than a little problem with submitting patches like this. We have no way to judge the suitability of the coding or the interface with the "meat" of the driver stubbed out!
This is also code for hardware that does not exist. In order for the code to be in distro releases in time for the hardware to ship, we must push it _before_ we have hardware. The main concern is getting in interfaces now, because the interfaces cannot change in a minor release. Providing more "meat" is on the ToDo list.
Signature · 2 lines
Russ Anderson, OS RAS/Partitioning Project Lead SGI - Silicon Graphics Inc [email protected]
H. Peter Anvin
reviewerQuoted 3 lines · Russ Anderson
/proc/sgi_uvalready exists, similar to/proc/sgi_snon Itanium systems.
Both those to be really awful.
Quoted 13 lines
Would it be
/sys/class/firmware? ReadingDocumentation/sysfs-rules.txtseems to indicate somewhere under/sys/devices, such as/sys/devices/system/, but I may be wrong.Quoted 8 lines
include/asm-x86/uv/bios.h struct uv_bios_retval { · line 66
Quoted 7 lines
include/asm-x86/uv/bios.h +7
-0+# define BIOS_CALL (result, a0, a1, a2, a3, a4, a5, a6, a7) \ + do { \ +/* XXX - the real call goes here */ \ + result.status =BIOS_STATUS_UNIMPLEMENTED ; \ + result.v0 =0 ; \ + result.v1 =0 ; \ + }while (0 )I have more than a little problem with submitting patches like this. We have no way to judge the suitability of the coding or the interface with the "meat" of the driver stubbed out!
Quoted 5 lines
This is also code for hardware that does not exist. In order for the code to be in distro releases in time for the hardware to ship, we must push it _before_ we have hardware. The main concern is getting in interfaces now, because the interfaces cannot change in a minor release. Providing more "meat" is on the ToDo list.
You could still document the intended interface so we know how it is supposed to work. At this point, you're basically saying "trust us", which really isn't how the Linux community is supposed to work.
-hpa
H. Peter Anvin
reviewerQuoted 7 lines · Russ Anderson
Quoted 1 line
This really seems more that it should belong in sysfs --
/sys/class/firmwarepresumably.
/proc/sgi_uvalready exists, similar to/proc/sgi_snon Itanium systems.Would it be
/sys/class/firmware? ReadingDocumentation/sysfs-rules.txtseems to indicate somewhere under/sys/devices, such as/sys/devices/system/, but I may be wrong.
/sys/class/firmware is usually where firmware-related things end up, but gregkh is the authority on sysfs organization.
-hpa
Quoted 11 lines · H. Peter Anvin
Quoted 7 lines · Russ Anderson
Quoted 2 lines
This really seems more that it should belong in sysfs --
/sys/class/firmwarepresumably.
/proc/sgi_uvalready exists, similar to/proc/sgi_snon Itanium systems. Would it be/sys/class/firmware? ReadingDocumentation/sysfs-rules.txtseems to indicate somewhere under/sys/devices, such as/sys/devices/system/,
but I may be wrong.
/sys/class/firmwareis usually where firmware-related things end up, but gregkh is the authority on sysfs organization.
/sys/class/firmware has a few users, but perhaps you are thinking of /sys/firmware instead? That's probably the better place for something that is not using the firmware class itself.
thanks,
greg k-h
Quoted 10 lines · Greg KH
Quoted 9 lines · H. Peter Anvin
Quoted 7 lines · Russ Anderson
Quoted 2 lines
This really seems more that it should belong in sysfs --
/sys/class/firmwarepresumably.
/proc/sgi_uvalready exists, similar to/proc/sgi_snon Itanium systems. Would it be/sys/class/firmware? ReadingDocumentation/sysfs-rules.txtseems to indicate somewhere under/sys/devices, such as/sys/devices/system/,
but I may be wrong.
Quoted 6 lines
Quoted 2 lines
/sys/class/firmwareis usually where firmware-related things end up, but gregkh is the authority on sysfs organization.
/sys/class/firmwarehas a few users, but perhaps you are thinking of/sys/firmwareinstead? That's probably the better place for something that is not using the firmware class itself.
Hi Greg,
If /sys/firmware is the right place, that is where I will put it.
Are the rules for /sys/firmware documented? Documentation/sysfs-rules.txt does not mention /sys/firmware. It seems to indicate that /sys/class should only contain symlinks:
There are currently three places for classification of devices:
/sys/block, /sys/class and /sys/bus. It is planned that these will not contain any device directories themselves, but only flat lists of symlinks pointing to the unified /sys/devices tree.
Thanks,
Signature · 2 lines
Russ Anderson, OS RAS/Partitioning Project Lead SGI - Silicon Graphics Inc [email protected]
Quoted 18 lines · Russ Anderson
Quoted 10 lines · Greg KH
Quoted 9 lines · H. Peter Anvin
Quoted 7 lines · Russ Anderson
Quoted 2 lines
This really seems more that it should belong in sysfs --
/sys/class/firmwarepresumably.
/proc/sgi_uvalready exists, similar to/proc/sgi_snon Itanium systems. Would it be/sys/class/firmware? ReadingDocumentation/sysfs-rules.txtseems to indicate somewhere under/sys/devices, such as/sys/devices/system/,
but I may be wrong.Quoted 6 lines
Quoted 2 lines
/sys/class/firmwareis usually where firmware-related things end up, but gregkh is the authority on sysfs organization.
/sys/class/firmwarehas a few users, but perhaps you are thinking of/sys/firmwareinstead? That's probably the better place for something that is not using the firmware class itself.
Quoted 7 lines
Hi Greg,
If
/sys/firmwareis the right place, that is where I will put it.Are the rules for
/sys/firmwaredocumented?Documentation/sysfs-rules.txtdoes not mention/sys/firmware. It seems to indicate that/sys/classshould only contain symlinks:
Yes, using /sys/class is when using a 'struct device', is that what you are using, or are you stuck with "raw" kobjects?
What exactly are you wanting to place in sysfs? That will make it easier to determine where to put it :)
Do you have a patch showing what you are proposing?
thanks,
greg k-h
Quoted 4 lines · Greg KH
Quoted 3 lines · Russ Anderson
Hi Greg,
Quoted 8 lines
Quoted 5 lines
If
/sys/firmwareis the right place, that is where I will put it.Are the rules for
/sys/firmwaredocumented?Documentation/sysfs-rules.txtdoes not mention/sys/firmware. It seems to indicate that/sys/classshould only contain symlinks:Yes, using
/sys/classis when using a 'struct device', is that what you are using, or are you stuck with "raw" kobjects?
Whatever makes the most sense.
Quoted 2 lines
What exactly are you wanting to place in sysfs? That will make it easier to determine where to put it :)
This patch has four entries: partition_id, coherence_id, system_serial_number, and licenseID. Each of the values comes from bios. The values are either u64 or char string. The values are read only in sysfs.
On ia64, this functionality is in /proc/sgi_sn/ for SGI Altix. The intent is to provide similar functionality for SGI UV on x86. ia64 linux uses SAL calls to get the information. For x86, the intent is to have comparable bios calls to get the data.
Quoted 1 line
Do you have a patch showing what you are proposing?
This is the patch using /proc/.
-----------------------------------------------------------------
SGI UV support for partitioning and system serial number.
--- arch/x86/kernel/Makefile | 2 arch/x86/kernel/bios_uv.c | 87 +++++++++++++++++++++++++- arch/x86/kernel/genx2apic_uv_x.c | 8 +- arch/x86/kernel/tlb_uv.c | 5 + arch/x86/kernel/uv_proc_fs.c | 122 +++++++++++++++++++++++++++++++++++++ include/asm-x86/uv/bios.h | 127 ++++++++++++++++++++++++++++++++++----- 6 files changed, 326 insertions(+), 25 deletions(-)
arch/x86/kernel/tlb_uv.c +4 -1
@@ -14,6 +14,7 @@#include <asm/uv/uv_mmrs.h> #include <asm/uv/uv_hub.h> #include <asm/uv/uv_bau.h> +#include <asm/uv/bios.h> #include <asm/genapic.h> #include <asm/idle.h> #include <asm/tsc.h>
@@ -565,7 +566,9 @@ static int __init uv_ptc_init(void)if (!is_uv_system ())return 0 ; -if (!proc_mkdir ("sgi_uv" ,NULL )) +if (!sgi_uv_dir) + sgi_uv_dir =proc_mkdir ("sgi_uv" ,NULL ); +if (!sgi_uv_dir)return -EINVAL ; proc_uv_ptc =create_proc_entry (UV_PTC_BASENAME ,0444 ,NULL );
arch/x86/kernel/uv_proc_fs.c new file +122 -0
122 lines not on this page · raw message
arch/x86/kernel/Makefile +1 -1
@@ -104,7 +104,7 @@ obj-$(CONFIG_OLPC) += olpc.o # 64 bit specific files ifeq ($(CONFIG_X86_64),y) obj-y += genapic_64.o genapic_flat_64.o genx2apic_uv_x.o tlb_uv.o - obj-y += bios_uv.o + obj-y += bios_uv.o uv_proc_fs.o obj-y += genx2apic_cluster.o obj-y += genx2apic_phys.o obj-$(CONFIG_X86_PM_TIMER) += pmtimer_64.o
arch/x86/kernel/bios_uv.c +82 -5
113 lines not on this page · raw message
arch/x86/kernel/genx2apic_uv_x.c +6 -2
@@ -344,7 +344,7 @@ static __init void uv_rtc_init(void)long status, ticks_per_sec, drift; status = -x86_bios_freq_base (BIOS_FREQ_BASE_REALTIME_CLOCK , &ticks_per_sec, +uv_bios_freq_base (BIOS_FREQ_BASE_REALTIME_CLOCK , &ticks_per_sec, &drift);if (status !=0 || ticks_per_sec <100000 ) { printk(KERN_WARNING
@@ -410,6 +410,10 @@ static __init void uv_system_init(void) gnode_upper = (((unsigned long )node_id.s.node_id) & ~((1 << n_val) -1 )) << m_val; +uv_bios_get_sn_info (0 , &uv_type, &uv_nasid_bitmask, + &uv_nasid_shift, &uv_system_size, + &uv_sharing_domain_size, &sn_partition_id, + &uv_coherency_id, &uv_region_size);uv_rtc_init ();for_each_present_cpu (cpu) {
@@ -431,7 +435,7 @@ static __init void uv_system_init(void)uv_cpu_hub_info (cpu)->gpa_mask = (1 << (m_val + n_val)) -1 ;uv_cpu_hub_info (cpu)->gnode_upper = gnode_upper;uv_cpu_hub_info (cpu)->global_mmr_base = mmr_base; -uv_cpu_hub_info (cpu)->coherency_domain_number =0 ;/* ZZZ */ +uv_cpu_hub_info (cpu)->coherency_domain_number = uv_coherency_id; uv_node_to_blade[nid] = blade; uv_cpu_to_blade[cpu] = blade; max_pnode =max (pnode, max_pnode);
include/asm-x86/uv/bios.h +111 -16
156 lines not on this page · raw message
Signature · 2 lines
Russ Anderson, OS RAS/Partitioning Project Lead SGI - Silicon Graphics Inc [email protected]
Quoted 13 lines · Greg KH
Quoted 11 lines · H. Peter Anvin
Quoted 7 lines · Russ Anderson
Quoted 2 lines
This really seems more that it should belong in sysfs --
/sys/class/firmwarepresumably.
/proc/sgi_uvalready exists, similar to/proc/sgi_snon Itanium systems. Would it be/sys/class/firmware? ReadingDocumentation/sysfs-rules.txtseems to indicate somewhere under/sys/devices, such as/sys/devices/system/,
but I may be wrong.
/sys/class/firmwareis usually where firmware-related things end up, but gregkh is the authority on sysfs organization.
Quoted 7 lines
/sys/class/firmwarehas a few users, but perhaps you are thinking of/sys/firmwareinstead? That's probably the better place for something that is not using the firmware class itself.thanks,
greg k-h
Here is an updated patch using /sys/firmware/sgi_uv/, to see if it resolves concerns in that area. This patch does not address the stubs issue. Still working that.
Thanks,
Russ
---------------------------------------------------------------------
SGI UV support for partitioning and system serial number.
--- arch/x86/kernel/Makefile | 2 arch/x86/kernel/bios_uv.c | 87 +++++++++++++++++++++++++- arch/x86/kernel/genx2apic_uv_x.c | 8 +- arch/x86/kernel/uv_sysfs.c | 84 ++++++++++++++++++++++++++ include/asm-x86/uv/bios.h | 126 ++++++++++++++++++++++++++++++++++----- 5 files changed, 283 insertions(+), 24 deletions(-)
arch/x86/kernel/uv_sysfs.c new file +84 -0
@@ -0,0 +1,84 @@ +/* +* This file supports the /sys/firmware/sgi_uv interfaces for SGI UV. +* +* This program is free software; you can redistribute it and/or modify +* it under the terms of the GNU General Public License as published by +* the Free Software Foundation; either version 2 of the License, or +* (at your option) any later version. +* +* This program is distributed in the hope that it will be useful, +* but WITHOUT ANY WARRANTY; without even the implied warranty of +* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +* GNU General Public License for more details. +* +* You should have received a copy of the GNU General Public License +* along with this program; if not, write to the Free Software +* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA +* +* Copyright (c) 2008 Silicon Graphics, Inc. All Rights Reserved. +* Copyright (c) Russ Anderson +*/ + +#include <linux/sysdev.h> +/* #include <linux/seq_file.h> */ +/* #include <asm/genapic.h> */ +#include <asm/uv/bios.h> + +struct kobject *sgi_uv_kobj; + +static ssize_t partition_id_show (struct kobject *kobj, +struct kobj_attribute *attr,char *buf) +{ +return snprintf (buf,PAGE_SIZE ,"%d\n" , sn_partition_id); +} + +static ssize_t system_serial_number_show (struct kobject *kobj, +struct kobj_attribute *attr,char *buf) +{ +return snprintf (buf,PAGE_SIZE ,"%s\n" ,sn_system_serial_number ()); +} + +static ssize_t licenseID_show (struct kobject *kobj, +struct kobj_attribute *attr,char *buf) +{ +return snprintf (buf,PAGE_SIZE ,"0x%llx\n" , +sn_partition_serial_number_val ()); +} + +static ssize_t coherence_id_show (struct kobject *kobj, +struct kobj_attribute *attr,char *buf) +{ +return snprintf (buf,PAGE_SIZE ,"%d\n" ,partition_coherence_id ()); +} + +static struct kobj_attribute partition_id_attr = +__ATTR (partition_id,S_IRUGO , partition_id_show,NULL ); + +static struct kobj_attribute system_sn_attr = +__ATTR (system_serial_number,S_IRUGO , system_serial_number_show,NULL ); + +static struct kobj_attribute license_id_attr = +__ATTR (licenseID,S_IRUGO , licenseID_show,NULL ); + +static struct kobj_attribute coherence_id_attr = +__ATTR (coherence_id,S_IRUGO , coherence_id_show,NULL ); + + +static int __init sgi_uv_sysfs_init (void ) +{ +if (!sgi_uv_kobj) + sgi_uv_kobj =kobject_create_and_add ("sgi_uv" , firmware_kobj); +if (!sgi_uv_kobj) { +printk (KERN_WARNING "kobject_create_and_add sgi_uv failed \n" ); +return -EINVAL ; + } + +sysfs_create_file (sgi_uv_kobj, &partition_id_attr.attr); +sysfs_create_file (sgi_uv_kobj, &system_sn_attr.attr); +sysfs_create_file (sgi_uv_kobj, &license_id_attr.attr); +sysfs_create_file (sgi_uv_kobj, &coherence_id_attr.attr); + +return 0 ; +} + +device_initcall (sgi_uv_sysfs_init);
arch/x86/kernel/Makefile +1 -1
@@ -104,7 +104,7 @@ obj-$(CONFIG_OLPC) += olpc.o # 64 bit specific files ifeq ($(CONFIG_X86_64),y) obj-y += genapic_64.o genapic_flat_64.o genx2apic_uv_x.o tlb_uv.o - obj-y += bios_uv.o + obj-y += bios_uv.o uv_sysfs.o obj-y += genx2apic_cluster.o obj-y += genx2apic_phys.o obj-$(CONFIG_X86_PM_TIMER) += pmtimer_64.o
arch/x86/kernel/bios_uv.c +82 -5
113 lines not on this page · raw message
arch/x86/kernel/genx2apic_uv_x.c +6 -2
@@ -344,7 +344,7 @@ static __init void uv_rtc_init(void)long status, ticks_per_sec, drift; status = -x86_bios_freq_base (BIOS_FREQ_BASE_REALTIME_CLOCK , &ticks_per_sec, +uv_bios_freq_base (BIOS_FREQ_BASE_REALTIME_CLOCK , &ticks_per_sec, &drift);if (status !=0 || ticks_per_sec <100000 ) { printk(KERN_WARNING
@@ -410,6 +410,10 @@ static __init void uv_system_init(void) gnode_upper = (((unsigned long )node_id.s.node_id) & ~((1 << n_val) -1 )) << m_val; +uv_bios_get_sn_info (0 , &uv_type, &uv_nasid_bitmask, + &uv_nasid_shift, &uv_system_size, + &uv_sharing_domain_size, &sn_partition_id, + &uv_coherency_id, &uv_region_size);uv_rtc_init ();for_each_present_cpu (cpu) {
@@ -431,7 +435,7 @@ static __init void uv_system_init(void)uv_cpu_hub_info (cpu)->gpa_mask = (1 << (m_val + n_val)) -1 ;uv_cpu_hub_info (cpu)->gnode_upper = gnode_upper;uv_cpu_hub_info (cpu)->global_mmr_base = mmr_base; -uv_cpu_hub_info (cpu)->coherency_domain_number =0 ;/* ZZZ */ +uv_cpu_hub_info (cpu)->coherency_domain_number = uv_coherency_id; uv_node_to_blade[nid] = blade; uv_cpu_to_blade[cpu] = blade; max_pnode =max (pnode, max_pnode);
include/asm-x86/uv/bios.h +110 -16
155 lines not on this page · raw message
Signature · 2 lines
Russ Anderson, OS RAS/Partitioning Project Lead SGI - Silicon Graphics Inc [email protected]
Quoted 14 lines · Russ Anderson
Quoted 9 lines · H. Peter Anvin
include/asm-x86/uv/bios.h struct uv_bios_retval { · line 66
Quoted 7 lines · Russ Anderson
include/asm-x86/uv/bios.h +7
-0+# define BIOS_CALL (result, a0, a1, a2, a3, a4, a5, a6, a7) \ + do { \ +/* XXX - the real call goes here */ \ + result.status =BIOS_STATUS_UNIMPLEMENTED ; \ + result.v0 =0 ; \ + result.v1 =0 ; \ + }while (0 )Quoted 3 lines
I have more than a little problem with submitting patches like this. We have no way to judge the suitability of the coding or the interface with the "meat" of the driver stubbed out!
Quoted 5 lines
This is also code for hardware that does not exist. In order for the code to be in distro releases in time for the hardware to ship, we must push it _before_ we have hardware. The main concern is getting in interfaces now, because the interfaces cannot change in a minor release. Providing more "meat" is on the ToDo list.
These kinds of statements are not the way to get things done on LKML. Quite a number of people have gotten flamed/ignored/etc lately by making statements "We need this (unfinished code) in the kernel by 2.6.27 because RHEL6 is going to use 2.6.27 and otherwise we won't be supported in RHEL6".
Cheers,
Kyle Moffett
Quoted 19 lines · Kyle Moffett
Quoted 9 lines · Russ Anderson
Quoted 8 lines · H. Peter Anvin
Quoted 7 lines · Russ Anderson
include/asm-x86/uv/bios.h +7
-0+# define BIOS_CALL (result, a0, a1, a2, a3, a4, a5, a6, a7) \ + do { \ +/* XXX - the real call goes here */ \ + result.status =BIOS_STATUS_UNIMPLEMENTED ; \ + result.v0 =0 ; \ + result.v1 =0 ; \ + }while (0 )Quoted 8 lines
Quoted 3 lines
I have more than a little problem with submitting patches like this. We have no way to judge the suitability of the coding or the interface with the "meat" of the driver stubbed out!
This is also code for hardware that does not exist. In order for the code to be in distro releases in time for the hardware to ship, we must push it _before_ we have hardware. The main concern is getting in interfaces now, because the interfaces cannot change in a minor release. Providing more "meat" is on the ToDo list.
Quoted 8 lines
These kinds of statements are not the way to get things done on LKML. Quite a number of people have gotten flamed/ignored/etc lately by making statements "We need this (unfinished code) in the kernel by 2.6.27 because RHEL6 is going to use 2.6.27 and otherwise we won't be supported in RHEL6".
Cheers,
Kyle Moffett
Hi Kyle,
As I'm very new to this development arena, could you explain a bit more on why this is considered "bad manners"?
I'm not speaking of any particular change, but there are some realities in bringing a new product to market that depends heavily on new "features" being accepted into a specific kernel release. I certainly do not want to "taint" any kernel code (and I'm always amazed at the dedication of so many individuals to insure this doesn't happen), but the line between acceptability (and not) seems to waver all over the place... ;-)
Thanks,
Mike
H. Peter Anvin
reviewerQuoted 6 lines · Mike Travis
Hi Kyle,
As I'm very new to this development arena, could you explain a bit more on why this is considered "bad manners"?
Quoted 7 lines
I'm not speaking of any particular change, but there are some realities in bringing a new product to market that depends heavily on new "features" being accepted into a specific kernel release. I certainly do not want to "taint" any kernel code (and I'm always amazed at the dedication of so many individuals to insure this doesn't happen), but the line between acceptability (and not) seems to waver all over the place... ;-)
It's because it's your responsibility to get the code in by whenever you need it to, but trying to push unfinished code with the motivation "we need it in by <release>" violates the development model *and* is just plain rude.
This comes down to the old saying "lack of planning on your part does not constitute an emergency on my part."
In other words, if you want to push code in by a specific release, the code needs to be *done* and properly submitted. Submitting code that has a big "real code goes here" comment, is ridiculous.
Unfortunately we have seen a *lot* of that from several people at SGI over the last year.
-hpa
Quoted 24 lines · H. Peter Anvin
Quoted 6 lines · Mike Travis
Hi Kyle,
As I'm very new to this development arena, could you explain a bit more on why this is considered "bad manners"?
Quoted 8 lines
I'm not speaking of any particular change, but there are some realities in
bringing a new product to market that depends heavily on new "features" being accepted into a specific kernel release. I certainly do not want to "taint" any kernel code (and I'm always amazed at the dedication of so many individuals to insure this doesn't happen), but the line between acceptability (and not) seems to waver all over the place... ;-)It's because it's your responsibility to get the code in by whenever you need it to, but trying to push unfinished code with the motivation "we need it in by <release>" violates the development model *and* is just plain rude.
This comes down to the old saying "lack of planning on your part does not constitute an emergency on my part."
Quoted 8 lines
In other words, if you want to push code in by a specific release, the code needs to be *done* and properly submitted. Submitting code that has a big "real code goes here" comment, is ridiculous.
Unfortunately we have seen a *lot* of that from several people at SGI over the last year.
-hpa
Hi Peter,
Ok, thanks, I do see your point (very clearly), as I prepare yet another "we really, really need this" patch... ;-)
[ok, it's only a led driver and the world won't stop if it doesn't show up in the kernel... ;-)]
Cheers,
Mike