Skip to contentSkip to conversation

arm64: kexec_file_load support

RFC AKASHI Takahiro 3 participants 12 messages Jul 1, 2016 – Jul 11, 2016

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Hi,

I'm not sure whether there is any demand for kexec_file_load support on arm64, but anyhow I'm working on this and now my early prototype code does work fine.

There is, however, one essential issue:
While arm64 kernel requires a device tree blob to be set up correctly at boot time, the current system call API doesn't have this parameter.
int kexec_file_load(int kernel_fd, int initrd_fd,
unsigned long cmdline_len, const char *cmdline_ptr, unsigned long flags);

Should we invent a new system call, like kexec_file_load2, and, if so, what kind of interface would be desired?

Thanks,
-Takahiro AKASHI

#1 lore raw

Am Freitag, 01 Juli 2016, 14:11:12 schrieb AKASHI Takahiro:

Quoted 3 lines

I'm not sure whether there is any demand for kexec_file_load support on arm64, but anyhow I'm working on this and now my early prototype code does work fine.

It is necessary if you want to support loading only signed kernels, and also if you want IMA to measure the kernel in its event log.

Quoted 10 lines

There is, however, one essential issue:
While arm64 kernel requires a device tree blob to be set up correctly at boot time, the current system call API doesn't have this parameter.
int kexec_file_load(int kernel_fd, int initrd_fd,
unsigned long cmdline_len, const char
*cmdline_ptr, unsigned long flags);

Should we invent a new system call, like kexec_file_load2, and, if so, what kind of interface would be desired?

I'm facing the same issue on powerpc. What I'm doing is taking the device tree that was used to boot the current kernel and modifying it as necessary to pass it to the next kernel.

I agree that it would be better if we could have a system call where a custom device tree could be passed. One suggestion is:

kexec_file_load2(int fds[], int fd_types[], int nr_fds,

		 unsigned long cmdline_len, const char *cmdline_ptr,
		unsigned long flags);

Where fds is an array with nr_fds file descriptors and fd_types is an array specifying what each fd in fds is. So for example, if fds[i] is the kernel, then fd_types[i] would have the value KEXEC_FILE_KERNEL_FD. If fds[i] is the device tree blob, fd_types[i], would have the value KEXEC_FILE_DTB and so on. That way, the syscall can be extended for an arbitrary number and types of segments that have to be loaded, just like kexec_load.

Another option is to have a struct:

kexec_file_load2(struct kexec_file_params *params, unsigned long params_sz);

Where:

struct kexec_file_params {
	int version;	/* allows struct to be extended in the future */
	int fds[];
	int fd_types[];
	int nr_fds;
	unsigned long cmdline_len;
	const char *cmdline_ptr;
	unsigned long flags;
};

This is even more flexible.

[]'s
Thiago Jung Bauermann
IBM Linux Technology Center

#2 lore raw

Hi,

Quoted 16 lines · Thiago Jung Bauermann

Am Freitag, 01 Juli 2016, 14:11:12 schrieb AKASHI Takahiro:

Quoted 3 lines

I'm not sure whether there is any demand for kexec_file_load support on arm64, but anyhow I'm working on this and now my early prototype code does work fine.

It is necessary if you want to support loading only signed kernels, and also if you want IMA to measure the kernel in its event log.

Quoted 8 lines

There is, however, one essential issue:
While arm64 kernel requires a device tree blob to be set up correctly at boot time, the current system call API doesn't have this parameter.
int kexec_file_load(int kernel_fd, int initrd_fd,
unsigned long cmdline_len, const char
*cmdline_ptr, unsigned long flags);

Quoted 6 lines
Quoted 2 lines

Should we invent a new system call, like kexec_file_load2, and, if so, what kind of interface would be desired?

I'm facing the same issue on powerpc. What I'm doing is taking the device tree that was used to boot the current kernel and modifying it as necessary to pass it to the next kernel.

That is exactly what I do.

Quoted 2 lines

I agree that it would be better if we could have a system call where a custom device tree could be passed. One suggestion is:

For powerpc, you might be able to use dtbImage instead of Image without changing the kernel interfaces.

Quoted 4 lines

kexec_file_load2(int fds[], int fd_types[], int nr_fds,

		 unsigned long cmdline_len, const char *cmdline_ptr,
		unsigned long flags);

You don't want to simply add one more argument, i.e. dtb_fd, don't you.

I prefer a slightly-simpler interface:

        struct kexec_file_fd {
                enum kexec_file_type;
                int fd;
        }

        int kexec_file_load2(struct kexec_file_fd[], int nr_fds, int flags);

Or if you want to keep the compatibility with the existing system call,

int kexec_file_load(int kernel_fd, int initrd_fd,
unsigned long cmdline_len, const char *cmdline_ptr, unsigned long flags,
int struct kexec_file_fd[], int nr_fds);

Here SYSCALL_DEFINE7() have to be defined, and I'm not sure that we will not have a problem in adding a system call with more than 6 arguments.

Quoted 10 lines

Where fds is an array with nr_fds file descriptors and fd_types is an array specifying what each fd in fds is. So for example, if fds[i] is the kernel, then fd_types[i] would have the value KEXEC_FILE_KERNEL_FD. If fds[i] is the device tree blob, fd_types[i], would have the value KEXEC_FILE_DTB and so on. That way, the syscall can be extended for an arbitrary number and types of segments that have to be loaded, just like kexec_load.

Another option is to have a struct:

kexec_file_load2(struct kexec_file_params *params, unsigned long params_sz);

Wow, we can add any number of new parameters with this interface.

Thanks,
-Takahiro AKASHI

Quoted 18 lines

Where:

struct kexec_file_params {
	int version;	/* allows struct to be extended in the future */
	int fds[];
	int fd_types[];
	int nr_fds;
	unsigned long cmdline_len;
	const char *cmdline_ptr;
	unsigned long flags;
};

This is even more flexible.

[]'s
Thiago Jung Bauermann
IBM Linux Technology Center

#3 lore raw

Hello,

Am Montag, 04 Juli 2016, 15:58:15 schrieb AKASHI Takahiro:

Quoted 6 lines · Thiago Jung Bauermann
Quoted 3 lines

I agree that it would be better if we could have a system call where a

custom device tree could be passed. One suggestion is:

For powerpc, you might be able to use dtbImage instead of Image without changing the kernel interfaces.

That works for custom kernels, but for signed kernels from a distro, I believe that's not an option.

Quoted 7 lines
Quoted 5 lines

kexec_file_load2(int fds[], int fd_types[], int nr_fds,

		 unsigned long cmdline_len, const char *cmdline_ptr,
		
		unsigned long flags);

You don't want to simply add one more argument, i.e. dtb_fd, don't you.

I'm just trying to avoid having to add another argument later if we find out someone is loading another segment that we didn't know about. :-)

The older kexec_load system call allows passing an arbitrary number of segments (sort of, currently capped at 16) to the kernel, so my suggestions preserve that feature.

If people think that adding another argument for dtb_fd is enough, I won't mind.

Quoted 8 lines

I prefer a slightly-simpler interface:

        struct kexec_file_fd {
                enum kexec_file_type;
                int fd;
        }

        int kexec_file_load2(struct kexec_file_fd[], int nr_fds, int

flags);

I like this one.

Quoted 9 lines

Or if you want to keep the compatibility with the existing system call,

int kexec_file_load(int kernel_fd, int initrd_fd,
unsigned long cmdline_len, const char
*cmdline_ptr, unsigned long flags,

                        int struct kexec_file_fd[], int nr_fds);

Here SYSCALL_DEFINE7() have to be defined, and I'm not sure that we will not have a problem in adding a system call with more than 6 arguments.

That's very clever. We can do what you suggest above or even just add dtb_fd with SYSCALL_DEFINE6. Either option would be good.

Quoted 8 lines
Quoted 8 lines

Where fds is an array with nr_fds file descriptors and fd_types is an array specifying what each fd in fds is. So for example, if fds[i] is the kernel, then fd_types[i] would have the value KEXEC_FILE_KERNEL_FD. If fds[i] is the device tree blob, fd_types[i], would have the value KEXEC_FILE_DTB and so on. That way, the syscall can be extended for an arbitrary number and types of segments that have to be loaded, just like kexec_load.

Quoted 5 lines
Quoted 4 lines

Another option is to have a struct:

kexec_file_load2(struct kexec_file_params *params, unsigned long params_sz);

Wow, we can add any number of new parameters with this interface.

Yeah, maybe it's a bit too much.

Signature · 3 lines
[]'s
Thiago Jung Bauermann
IBM Linux Technology Center
#5 lore raw
In reply to AKASHI Takahiro
Quoted 65 lines · AKASHI Takahiro

Hi,

Quoted 16 lines · Thiago Jung Bauermann

Am Freitag, 01 Juli 2016, 14:11:12 schrieb AKASHI Takahiro:

Quoted 3 lines

I'm not sure whether there is any demand for kexec_file_load support on arm64, but anyhow I'm working on this and now my early prototype code does work fine.

It is necessary if you want to support loading only signed kernels, and also if you want IMA to measure the kernel in its event log.

Quoted 8 lines

There is, however, one essential issue:
While arm64 kernel requires a device tree blob to be set up correctly at boot time, the current system call API doesn't have this parameter.
int kexec_file_load(int kernel_fd, int initrd_fd,
unsigned long cmdline_len, const char
*cmdline_ptr, unsigned long flags);

Quoted 6 lines
Quoted 2 lines

Should we invent a new system call, like kexec_file_load2, and, if so, what kind of interface would be desired?

I'm facing the same issue on powerpc. What I'm doing is taking the device tree that was used to boot the current kernel and modifying it as necessary to pass it to the next kernel.

That is exactly what I do.

Quoted 2 lines

I agree that it would be better if we could have a system call where a custom device tree could be passed. One suggestion is:

For powerpc, you might be able to use dtbImage instead of Image without changing the kernel interfaces.

Quoted 4 lines

kexec_file_load2(int fds[], int fd_types[], int nr_fds,

		 unsigned long cmdline_len, const char *cmdline_ptr,
		unsigned long flags);

You don't want to simply add one more argument, i.e. dtb_fd, don't you.

I prefer a slightly-simpler interface:

        struct kexec_file_fd {
                enum kexec_file_type;
                int fd;
        }

        int kexec_file_load2(struct kexec_file_fd[], int nr_fds, int flags);

Or if you want to keep the compatibility with the existing system call,

int kexec_file_load(int kernel_fd, int initrd_fd,
unsigned long cmdline_len, const char *cmdline_ptr, unsigned long flags,
int struct kexec_file_fd[], int nr_fds);

Here SYSCALL_DEFINE7() have to be defined, and I'm not sure that we will not have a problem in adding a system call with more than 6 arguments.

Quoted 7 lines

Where fds is an array with nr_fds file descriptors and fd_types is an array specifying what each fd in fds is. So for example, if fds[i] is the kernel, then fd_types[i] would have the value KEXEC_FILE_KERNEL_FD. If fds[i] is the device tree blob, fd_types[i], would have the value KEXEC_FILE_DTB and so on. That way, the syscall can be extended for an arbitrary number and types of segments that have to be loaded, just like kexec_load.

Quoted 22 lines
Quoted 3 lines

Another option is to have a struct:

kexec_file_load2(struct kexec_file_params *params, unsigned long params_sz);

Wow, we can add any number of new parameters with this interface.

Thanks,
-Takahiro AKASHI

Quoted 13 lines

Where:

struct kexec_file_params {
	int version;	/* allows struct to be extended in the future */
	int fds[];
	int fd_types[];
	int nr_fds;
	unsigned long cmdline_len;
	const char *cmdline_ptr;
	unsigned long flags;
};

This is even more flexible.

I would like to vote for this one, and use kexec_file_fd fds[] in the struct

Thanks
Dave

_______________________________________________
kexec mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/kexec

#6 lore raw

Hi Dave,

Quoted 87 lines · Dave Young
Quoted 86 lines · AKASHI Takahiro

Hi,

Quoted 16 lines · Thiago Jung Bauermann

Am Freitag, 01 Juli 2016, 14:11:12 schrieb AKASHI Takahiro:

Quoted 3 lines

I'm not sure whether there is any demand for kexec_file_load support on arm64, but anyhow I'm working on this and now my early prototype code does work fine.

It is necessary if you want to support loading only signed kernels, and also if you want IMA to measure the kernel in its event log.

Quoted 8 lines

There is, however, one essential issue:
While arm64 kernel requires a device tree blob to be set up correctly at boot time, the current system call API doesn't have this parameter.
int kexec_file_load(int kernel_fd, int initrd_fd,
unsigned long cmdline_len, const char
*cmdline_ptr, unsigned long flags);

Quoted 6 lines
Quoted 2 lines

Should we invent a new system call, like kexec_file_load2, and, if so, what kind of interface would be desired?

I'm facing the same issue on powerpc. What I'm doing is taking the device tree that was used to boot the current kernel and modifying it as necessary to pass it to the next kernel.

That is exactly what I do.

Quoted 2 lines

I agree that it would be better if we could have a system call where a custom device tree could be passed. One suggestion is:

For powerpc, you might be able to use dtbImage instead of Image without changing the kernel interfaces.

Quoted 4 lines

kexec_file_load2(int fds[], int fd_types[], int nr_fds,

		 unsigned long cmdline_len, const char *cmdline_ptr,
		unsigned long flags);

You don't want to simply add one more argument, i.e. dtb_fd, don't you.

I prefer a slightly-simpler interface:

        struct kexec_file_fd {
                enum kexec_file_type;
                int fd;
        }

        int kexec_file_load2(struct kexec_file_fd[], int nr_fds, int flags);

Or if you want to keep the compatibility with the existing system call,

int kexec_file_load(int kernel_fd, int initrd_fd,
unsigned long cmdline_len, const char *cmdline_ptr, unsigned long flags,
int struct kexec_file_fd[], int nr_fds);

Here SYSCALL_DEFINE7() have to be defined, and I'm not sure that we will not have a problem in adding a system call with more than 6 arguments.

Quoted 10 lines

Where fds is an array with nr_fds file descriptors and fd_types is an array specifying what each fd in fds is. So for example, if fds[i] is the kernel, then fd_types[i] would have the value KEXEC_FILE_KERNEL_FD. If fds[i] is the device tree blob, fd_types[i], would have the value KEXEC_FILE_DTB and so on. That way, the syscall can be extended for an arbitrary number and types of segments that have to be loaded, just like kexec_load.

Another option is to have a struct:

kexec_file_load2(struct kexec_file_params *params, unsigned long params_sz);

Wow, we can add any number of new parameters with this interface.

Thanks,
-Takahiro AKASHI

Quoted 12 lines

Where:

struct kexec_file_params {
	int version;	/* allows struct to be extended in the future */
	int fds[];
	int fd_types[];
	int nr_fds;
	unsigned long cmdline_len;
	const char *cmdline_ptr;
	unsigned long flags;
};
Quoted 3 lines
Quoted 1 line
Quoted 1 line

This is even more flexible.

I would like to vote for this one, and use kexec_file_fd fds[] in the struct

If we take this approach, we'd better take "flags" out of struct, and my preference would be:

        enum kexec_file_type {
                KEXEC_FILE_TYPE_KERNEL;
                KEXEC_FILE_TYPE_INITRD;
                KEXEC_FILE_TYPE_DTB;
        }

        struct kexec_file_fd {
                enum kexec_file_type;
                int fd;
        }

        sturct kexec_file_params {
                int version;
                unsigned char *cmdline;
                unsigned long cmdline_len;
                int nr_fds;
                struct kexec_file_fd fds[0];
        }

        int kexec_file_load2(int kernel_fd, unsigned long flags,
                                sturct kexec_file_params extra);

So we don't have to retrieve extra if KEXEC_FILE_UNLOAD (or kernel_fd < 0?),
and only once retrieve extra if extra != NULL && nr_fds == 0.

Thanks,
-Takahiro AKASHI

Quoted 2 lines

Thanks
Dave

#4 lore raw
In reply to Thiago Jung Bauermann
Quoted 24 lines · Thiago Jung Bauermann

Hello,

Am Montag, 04 Juli 2016, 15:58:15 schrieb AKASHI Takahiro:

Quoted 6 lines · Thiago Jung Bauermann
Quoted 3 lines

I agree that it would be better if we could have a system call where a

custom device tree could be passed. One suggestion is:

For powerpc, you might be able to use dtbImage instead of Image without changing the kernel interfaces.

That works for custom kernels, but for signed kernels from a distro, I believe that's not an option.

Quoted 7 lines
Quoted 5 lines

kexec_file_load2(int fds[], int fd_types[], int nr_fds,

		 unsigned long cmdline_len, const char *cmdline_ptr,
		
		unsigned long flags);

You don't want to simply add one more argument, i.e. dtb_fd, don't you.

I'm just trying to avoid having to add another argument later if we find out someone is loading another segment that we didn't know about. :-)

Quoted 6 lines

The older kexec_load system call allows passing an arbitrary number of segments (sort of, currently capped at 16) to the kernel, so my suggestions preserve that feature.

If people think that adding another argument for dtb_fd is enough, I won't mind.

That is the question :)
As far as I look though existing arch code, there will be no extra parameters (segments) required other than dtb.

Thanks,
-Takahiro AKASHI

Quoted 45 lines
Quoted 8 lines

I prefer a slightly-simpler interface:

        struct kexec_file_fd {
                enum kexec_file_type;
                int fd;
        }

        int kexec_file_load2(struct kexec_file_fd[], int nr_fds, int

flags);

I like this one.

Quoted 9 lines

Or if you want to keep the compatibility with the existing system call,

int kexec_file_load(int kernel_fd, int initrd_fd,
unsigned long cmdline_len, const char
*cmdline_ptr, unsigned long flags,

                        int struct kexec_file_fd[], int nr_fds);

Here SYSCALL_DEFINE7() have to be defined, and I'm not sure that we will not have a problem in adding a system call with more than 6 arguments.

That's very clever. We can do what you suggest above or even just add dtb_fd with SYSCALL_DEFINE6. Either option would be good.

Quoted 8 lines
Quoted 8 lines

Where fds is an array with nr_fds file descriptors and fd_types is an array specifying what each fd in fds is. So for example, if fds[i] is the kernel, then fd_types[i] would have the value KEXEC_FILE_KERNEL_FD. If fds[i] is the device tree blob, fd_types[i], would have the value KEXEC_FILE_DTB and so on. That way, the syscall can be extended for an arbitrary number and types of segments that have to be loaded, just like kexec_load.

Quoted 5 lines
Quoted 4 lines

Another option is to have a struct:

kexec_file_load2(struct kexec_file_params *params, unsigned long params_sz);

Wow, we can add any number of new parameters with this interface.

Yeah, maybe it's a bit too much.

Signature · 3 lines
[]'s
Thiago Jung Bauermann
IBM Linux Technology Center
#7 lore raw
In reply to AKASHI Takahiro
Quoted 119 lines · AKASHI Takahiro

Hi Dave,

Quoted 87 lines · Dave Young
Quoted 86 lines · AKASHI Takahiro

Hi,

Quoted 16 lines · Thiago Jung Bauermann

Am Freitag, 01 Juli 2016, 14:11:12 schrieb AKASHI Takahiro:

Quoted 3 lines

I'm not sure whether there is any demand for kexec_file_load support on arm64, but anyhow I'm working on this and now my early prototype code does work fine.

It is necessary if you want to support loading only signed kernels, and also if you want IMA to measure the kernel in its event log.

Quoted 8 lines

There is, however, one essential issue:
While arm64 kernel requires a device tree blob to be set up correctly at boot time, the current system call API doesn't have this parameter.
int kexec_file_load(int kernel_fd, int initrd_fd,
unsigned long cmdline_len, const char
*cmdline_ptr, unsigned long flags);

Quoted 6 lines
Quoted 2 lines

Should we invent a new system call, like kexec_file_load2, and, if so, what kind of interface would be desired?

I'm facing the same issue on powerpc. What I'm doing is taking the device tree that was used to boot the current kernel and modifying it as necessary to pass it to the next kernel.

That is exactly what I do.

Quoted 2 lines

I agree that it would be better if we could have a system call where a custom device tree could be passed. One suggestion is:

For powerpc, you might be able to use dtbImage instead of Image without changing the kernel interfaces.

Quoted 4 lines

kexec_file_load2(int fds[], int fd_types[], int nr_fds,

		 unsigned long cmdline_len, const char *cmdline_ptr,
		unsigned long flags);

You don't want to simply add one more argument, i.e. dtb_fd, don't you.

I prefer a slightly-simpler interface:

        struct kexec_file_fd {
                enum kexec_file_type;
                int fd;
        }

        int kexec_file_load2(struct kexec_file_fd[], int nr_fds, int flags);

Or if you want to keep the compatibility with the existing system call,

int kexec_file_load(int kernel_fd, int initrd_fd,
unsigned long cmdline_len, const char *cmdline_ptr, unsigned long flags,
int struct kexec_file_fd[], int nr_fds);

Here SYSCALL_DEFINE7() have to be defined, and I'm not sure that we will not have a problem in adding a system call with more than 6 arguments.

Quoted 10 lines

Where fds is an array with nr_fds file descriptors and fd_types is an array specifying what each fd in fds is. So for example, if fds[i] is the kernel, then fd_types[i] would have the value KEXEC_FILE_KERNEL_FD. If fds[i] is the device tree blob, fd_types[i], would have the value KEXEC_FILE_DTB and so on. That way, the syscall can be extended for an arbitrary number and types of segments that have to be loaded, just like kexec_load.

Another option is to have a struct:

kexec_file_load2(struct kexec_file_params *params, unsigned long params_sz);

Wow, we can add any number of new parameters with this interface.

Thanks,
-Takahiro AKASHI

Quoted 12 lines

Where:

struct kexec_file_params {
	int version;	/* allows struct to be extended in the future */
	int fds[];
	int fd_types[];
	int nr_fds;
	unsigned long cmdline_len;
	const char *cmdline_ptr;
	unsigned long flags;
};
Quoted 3 lines
Quoted 1 line
Quoted 1 line

This is even more flexible.

I would like to vote for this one, and use kexec_file_fd fds[] in the struct

If we take this approach, we'd better take "flags" out of struct, and my preference would be:

        enum kexec_file_type {
                KEXEC_FILE_TYPE_KERNEL;
                KEXEC_FILE_TYPE_INITRD;
                KEXEC_FILE_TYPE_DTB;
        }

        struct kexec_file_fd {
                enum kexec_file_type;
                int fd;
        }

        sturct kexec_file_params {
                int version;
                unsigned char *cmdline;
                unsigned long cmdline_len;
                int nr_fds;
                struct kexec_file_fd fds[0];
        }

        int kexec_file_load2(int kernel_fd, unsigned long flags,
                                sturct kexec_file_params extra);
Quoted 3 lines

So we don't have to retrieve extra if KEXEC_FILE_UNLOAD (or kernel_fd < 0?),
and only once retrieve extra if extra != NULL && nr_fds == 0.

If so maybe change a bit from your precious mentioned 7 args proposal like below?

struct kexec_file_fd {
	enum kexec_file_type;
	int fd;
}
struct kexec_fdset {
	int nr_fd;
	struct kexec_file_fd fd[0];
}

int kexec_file_load(int kernel_fd, int initrd_fd,

		    unsigned long cmdline_len, const char *cmdline_ptr,
		    unsigned long flags, struct kexec_fdset *extra_fds);

Thanks
Dave

_______________________________________________
kexec mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/kexec

#8 lore raw

Am Donnerstag, 07 Juli 2016, 14:12:45 schrieb Dave Young:

Quoted 8 lines

If so maybe change a bit from your precious mentioned 7 args proposal like below?

struct kexec_file_fd {
	enum kexec_file_type;
	int fd;
}
Quoted 8 lines
struct kexec_fdset {
	int nr_fd;
	struct kexec_file_fd fd[0];
}

int kexec_file_load(int kernel_fd, int initrd_fd,

		    unsigned long cmdline_len, const char *cmdline_ptr,
		    unsigned long flags, struct kexec_fdset *extra_fds);

Is there a way for the kernel to distinguish whether the process passed 5 or 6 arguments? How can it know whether extra_fds is a valid argument or just garbage? I think we have to define a new flag KEXEC_FILE_EXTRA_FDS so that the process can signal that it is using the new interface.

Signature · 3 lines
[]'s
Thiago Jung Bauermann
IBM Linux Technology Center
#9 lore raw
Quoted 19 lines · Thiago Jung Bauermann

Am Donnerstag, 07 Juli 2016, 14:12:45 schrieb Dave Young:

Quoted 8 lines

If so maybe change a bit from your precious mentioned 7 args proposal like below?

struct kexec_file_fd {
	enum kexec_file_type;
	int fd;
}
Quoted 8 lines
struct kexec_fdset {
	int nr_fd;
	struct kexec_file_fd fd[0];
}

int kexec_file_load(int kernel_fd, int initrd_fd,

		    unsigned long cmdline_len, const char *cmdline_ptr,
		    unsigned long flags, struct kexec_fdset *extra_fds);
Quoted 4 lines

Is there a way for the kernel to distinguish whether the process passed 5 or 6 arguments? How can it know whether extra_fds is a valid argument or just garbage? I think we have to define a new flag KEXEC_FILE_EXTRA_FDS so that the process can signal that it is using the new interface.

Agreed, a new flag is needed.

Thanks
Dave

_______________________________________________
kexec mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/kexec

#10 lore raw
In reply to Thiago Jung Bauermann
Quoted 19 lines · Thiago Jung Bauermann

Am Donnerstag, 07 Juli 2016, 14:12:45 schrieb Dave Young:

Quoted 8 lines

If so maybe change a bit from your precious mentioned 7 args proposal like below?

struct kexec_file_fd {
	enum kexec_file_type;
	int fd;
}
Quoted 8 lines
struct kexec_fdset {
	int nr_fd;
	struct kexec_file_fd fd[0];
}

int kexec_file_load(int kernel_fd, int initrd_fd,

		    unsigned long cmdline_len, const char *cmdline_ptr,
		    unsigned long flags, struct kexec_fdset *extra_fds);
Quoted 4 lines

Is there a way for the kernel to distinguish whether the process passed 5 or 6 arguments? How can it know whether extra_fds is a valid argument or just garbage? I think we have to define a new flag KEXEC_FILE_EXTRA_FDS so that the process can signal that it is using the new interface.

Good point. What I had in my mind is "open" system call. Glibc's implementation of open() checks for the second argument, flags, to determine whether or not the third argument, mode, is required.

So our current consensus is that, for the time being, we will extend the existing system call interface to allow extra file descriptors, and *if* we really need a different type of parameters in the future, we will add another system call.

If we all agree, I can post a kernel patch for this change as a RFC. (I've already verified it with my prototype of kexec_file_load support on arm64.)

Thanks,
-Takahiro AKASHI

Quoted 5 lines
Signature · 3 lines
[]'s
Thiago Jung Bauermann
IBM Linux Technology Center
#11 lore raw
Quoted 25 lines · AKASHI Takahiro
Quoted 19 lines · Thiago Jung Bauermann

Am Donnerstag, 07 Juli 2016, 14:12:45 schrieb Dave Young:

Quoted 8 lines

If so maybe change a bit from your precious mentioned 7 args proposal like below?

struct kexec_file_fd {
	enum kexec_file_type;
	int fd;
}
Quoted 8 lines
struct kexec_fdset {
	int nr_fd;
	struct kexec_file_fd fd[0];
}

int kexec_file_load(int kernel_fd, int initrd_fd,

		    unsigned long cmdline_len, const char *cmdline_ptr,
		    unsigned long flags, struct kexec_fdset *extra_fds);
Quoted 4 lines

Is there a way for the kernel to distinguish whether the process passed 5 or 6 arguments? How can it know whether extra_fds is a valid argument or just garbage? I think we have to define a new flag KEXEC_FILE_EXTRA_FDS so that the process can signal that it is using the new interface.

Quoted 8 lines

Good point. What I had in my mind is "open" system call. Glibc's implementation of open() checks for the second argument, flags, to determine whether or not the third argument, mode, is required.

So our current consensus is that, for the time being, we will extend the existing system call interface to allow extra file descriptors, and *if* we really need a different type of parameters in the future, we will add another system call.

Sounds good to me.

Quoted 13 lines

If we all agree, I can post a kernel patch for this change as a RFC. (I've already verified it with my prototype of kexec_file_load support on arm64.)

Thanks,
-Takahiro AKASHI

Quoted 5 lines
Signature · 3 lines
[]'s
Thiago Jung Bauermann
IBM Linux Technology Center

Thanks
Dave

_______________________________________________
kexec mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/kexec