Date:   Thu, 5 Nov 2020 08:38:53 -0600
From:   Rob Herring <robh@...nel.org>
To:     Lakshmi Ramasubramanian <nramas@...ux.microsoft.com>
Cc:     Mimi Zohar <zohar@...ux.ibm.com>,
        Thiago Jung Bauermann <bauerman@...ux.ibm.com>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        James Morse <james.morse@....com>,
        Catalin Marinas <catalin.marinas@....com>,
        Sasha Levin <sashal@...nel.org>, Will Deacon <will@...nel.org>,
        Michael Ellerman <mpe@...erman.id.au>,
        Benjamin Herrenschmidt <benh@...nel.crashing.org>,
        Paul Mackerras <paulus@...ba.org>,
        Frank Rowand <frowand.list@...il.com>,
        vincenzo.frascino@....com, Mark Rutland <mark.rutland@....com>,
        dmitry.kasatkin@...il.com, James Morris <jmorris@...ei.org>,
        "Serge E. Hallyn" <serge@...lyn.com>,
        Pavel Tatashin <pasha.tatashin@...een.com>,
        Allison Randal <allison@...utok.net>,
        Kate Stewart <kstewart@...uxfoundation.org>,
        "AKASHI, Takahiro" <takahiro.akashi@...aro.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Masahiro Yamada <masahiroy@...nel.org>,
        Bhupesh Sharma <bhsharma@...hat.com>,
        Matthias Brugger <mbrugger@...e.com>,
        Hsin-Yi Wang <hsinyi@...omium.org>, tao.li@...o.com,
        Christophe Leroy <christophe.leroy@....fr>,
        linux-integrity@...r.kernel.org,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        devicetree@...r.kernel.org,
        Prakhar Srivastava <prsriva@...ux.microsoft.com>,
        balajib@...ux.microsoft.com
Subject: Re: [PATCH v8 1/4] powerpc: Refactor kexec functions to move arch
 independent code to drivers/of

On Wed, Nov 4, 2020 at 6:46 PM Lakshmi Ramasubramanian
<nramas@...ux.microsoft.com> wrote:
>
> On 11/4/20 2:28 PM, Rob Herring wrote:
>
> Hi Rob,
>
> Thanks for reviewing the patch.
>
> > On Fri, Oct 30, 2020 at 10:44:26AM -0700, Lakshmi Ramasubramanian wrote:
> >> The functions remove_ima_buffer() and delete_fdt_mem_rsv() that handle
> >> carrying forward the IMA measurement logs on kexec for powerpc do not
> >> have architecture specific code, but they are currently defined for
> >> powerpc only.
> >>
> >> remove_ima_buffer() and delete_fdt_mem_rsv() are used to remove
> >> the IMA log entry from the device tree and free the memory reserved
> >> for the log. These functions need to be defined even if the current
> >> kernel does not support carrying forward IMA log across kexec since
> >> the previous kernel could have supported that and therefore the current
> >> kernel needs to free the allocation.
> >>
> >> Rename remove_ima_buffer() to remove_ima_kexec_buffer().
> >> Define remove_ima_kexec_buffer() and delete_fdt_mem_rsv() in
> >> drivers/of/fdt.c. A later patch in this series will use these functions
> >> to free the allocation, if any, made by the previous kernel for ARM64.
> >>
> >> Define FDT_PROP_IMA_KEXEC_BUFFER for the chosen node, namely
> >> "linux,ima-kexec-buffer", that is added to the DTB to hold
> >> the address and the size of the memory reserved to carry
> >> the IMA measurement log.
> >>
> >> Co-developed-by: Prakhar Srivastava <prsriva@...ux.microsoft.com>
> >> Signed-off-by: Prakhar Srivastava <prsriva@...ux.microsoft.com>
> >> Signed-off-by: Lakshmi Ramasubramanian <nramas@...ux.microsoft.com>
> >> Reported-by: kernel test robot <lkp@...el.com> error: kernel/kexec_file_fdt.c:30: undefined reference to `fdt_num_mem_rsv'
> >
> > This should be added for a commit fixing the reported problem. 0-day
> > didn't report what this patch implements.
> I'll remove that in the updated patch I'll post.
>
> >
> >> ---
> >>   arch/powerpc/include/asm/ima.h   |  10 +--
> >>   arch/powerpc/include/asm/kexec.h |   1 -
> >>   arch/powerpc/kexec/file_load.c   |  33 +---------
> >>   arch/powerpc/kexec/ima.c         |  55 +++-------------
> >>   drivers/of/fdt.c                 | 110 +++++++++++++++++++++++++++++++
> >>   include/linux/kexec.h            |  24 +++++++
> >>   include/linux/libfdt.h           |   3 +
> >>   7 files changed, 149 insertions(+), 87 deletions(-)
> >>
> >> diff --git a/arch/powerpc/include/asm/ima.h b/arch/powerpc/include/asm/ima.h
> >> index ead488cf3981..6355a85a3289 100644
> >> --- a/arch/powerpc/include/asm/ima.h
> >> +++ b/arch/powerpc/include/asm/ima.h
> >> @@ -2,17 +2,13 @@
> >>   #ifndef _ASM_POWERPC_IMA_H
> >>   #define _ASM_POWERPC_IMA_H
> >>
> >> +#include <linux/kexec.h>
> >> +
> >>   struct kimage;
> >>
> >>   int ima_get_kexec_buffer(void **addr, size_t *size);
> >>   int ima_free_kexec_buffer(void);
> >>
> >> -#ifdef CONFIG_IMA
> >> -void remove_ima_buffer(void *fdt, int chosen_node);
> >> -#else
> >> -static inline void remove_ima_buffer(void *fdt, int chosen_node) {}
> >> -#endif
> >> -
> >>   #ifdef CONFIG_IMA_KEXEC
> >>   int arch_ima_add_kexec_buffer(struct kimage *image, unsigned long load_addr,
> >>                            size_t size);
> >> @@ -22,7 +18,7 @@ int setup_ima_buffer(const struct kimage *image, void *fdt, int chosen_node);
> >>   static inline int setup_ima_buffer(const struct kimage *image, void *fdt,
> >>                                 int chosen_node)
> >>   {
> >> -    remove_ima_buffer(fdt, chosen_node);
> >> +    remove_ima_kexec_buffer(fdt, chosen_node);
> >>      return 0;
> >>   }
> >>   #endif /* CONFIG_IMA_KEXEC */
> >> diff --git a/arch/powerpc/include/asm/kexec.h b/arch/powerpc/include/asm/kexec.h
> >> index 55d6ede30c19..7c223031ecdd 100644
> >> --- a/arch/powerpc/include/asm/kexec.h
> >> +++ b/arch/powerpc/include/asm/kexec.h
> >> @@ -126,7 +126,6 @@ int setup_purgatory(struct kimage *image, const void *slave_code,
> >>   int setup_new_fdt(const struct kimage *image, void *fdt,
> >>                unsigned long initrd_load_addr, unsigned long initrd_len,
> >>                const char *cmdline);
> >> -int delete_fdt_mem_rsv(void *fdt, unsigned long start, unsigned long size);
> >>
> >>   #ifdef CONFIG_PPC64
> >>   struct kexec_buf;
> >> diff --git a/arch/powerpc/kexec/file_load.c b/arch/powerpc/kexec/file_load.c
> >> index 9a232bc36c8f..7a17655c530e 100644
> >> --- a/arch/powerpc/kexec/file_load.c
> >> +++ b/arch/powerpc/kexec/file_load.c
> >> @@ -18,6 +18,7 @@
> >>   #include <linux/kexec.h>
> >>   #include <linux/of_fdt.h>
> >>   #include <linux/libfdt.h>
> >> +#include <linux/kexec.h>
> >>   #include <asm/setup.h>
> >>   #include <asm/ima.h>
> >>
> >> @@ -109,38 +110,6 @@ int setup_purgatory(struct kimage *image, const void *slave_code,
> >>      return 0;
> >>   }
> >>
> >> -/**
> >> - * delete_fdt_mem_rsv - delete memory reservation with given address and size
> >> - *
> >> - * Return: 0 on success, or negative errno on error.
> >> - */
> >> -int delete_fdt_mem_rsv(void *fdt, unsigned long start, unsigned long size)
> >> -{
> >> -    int i, ret, num_rsvs = fdt_num_mem_rsv(fdt);
> >> -
> >> -    for (i = 0; i < num_rsvs; i++) {
> >> -            uint64_t rsv_start, rsv_size;
> >> -
> >> -            ret = fdt_get_mem_rsv(fdt, i, &rsv_start, &rsv_size);
> >> -            if (ret) {
> >> -                    pr_err("Malformed device tree.\n");
> >> -                    return -EINVAL;
> >> -            }
> >> -
> >> -            if (rsv_start == start && rsv_size == size) {
> >> -                    ret = fdt_del_mem_rsv(fdt, i);
> >> -                    if (ret) {
> >> -                            pr_err("Error deleting device tree reservation.\n");
> >> -                            return -EINVAL;
> >> -                    }
> >> -
> >> -                    return 0;
> >> -            }
> >> -    }
> >> -
> >> -    return -ENOENT;
> >> -}
> >> -
> >>   /*
> >>    * setup_new_fdt - modify /chosen and memory reservation for the next kernel
> >>    * @image:         kexec image being loaded.
> >> diff --git a/arch/powerpc/kexec/ima.c b/arch/powerpc/kexec/ima.c
> >> index 720e50e490b6..2b790230ea15 100644
> >> --- a/arch/powerpc/kexec/ima.c
> >> +++ b/arch/powerpc/kexec/ima.c
> >> @@ -11,6 +11,8 @@
> >>   #include <linux/of.h>
> >>   #include <linux/memblock.h>
> >>   #include <linux/libfdt.h>
> >> +#include <linux/ima.h>
> >> +#include <asm/ima.h>
> >>
> >>   static int get_addr_size_cells(int *addr_cells, int *size_cells)
> >>   {
> >> @@ -28,24 +30,6 @@ static int get_addr_size_cells(int *addr_cells, int *size_cells)
> >>      return 0;
> >>   }
> >>
> >> -static int do_get_kexec_buffer(const void *prop, int len, unsigned long *addr,
> >> -                           size_t *size)
> >> -{
> >> -    int ret, addr_cells, size_cells;
> >> -
> >> -    ret = get_addr_size_cells(&addr_cells, &size_cells);
> >> -    if (ret)
> >> -            return ret;
> >> -
> >> -    if (len < 4 * (addr_cells + size_cells))
> >> -            return -ENOENT;
> >> -
> >> -    *addr = of_read_number(prop, addr_cells);
> >> -    *size = of_read_number(prop + 4 * addr_cells, size_cells);
> >> -
> >> -    return 0;
> >> -}
> >> -
> >>   /**
> >>    * ima_get_kexec_buffer - get IMA buffer from the previous kernel
> >>    * @addr:  On successful return, set to point to the buffer contents.
> >> @@ -100,37 +84,14 @@ int ima_free_kexec_buffer(void)
> >>
> >>   }
> >>
> >> -/**
> >> - * remove_ima_buffer - remove the IMA buffer property and reservation from @fdt
> >> - *
> >> - * The IMA measurement buffer is of no use to a subsequent kernel, so we always
> >> - * remove it from the device tree.
> >> - */
> >> -void remove_ima_buffer(void *fdt, int chosen_node)
> >> -{
> >> -    int ret, len;
> >> -    unsigned long addr;
> >> -    size_t size;
> >> -    const void *prop;
> >> -
> >> -    prop = fdt_getprop(fdt, chosen_node, "linux,ima-kexec-buffer", &len);
> >> -    if (!prop)
> >> -            return;
> >> -
> >> -    ret = do_get_kexec_buffer(prop, len, &addr, &size);
> >> -    fdt_delprop(fdt, chosen_node, "linux,ima-kexec-buffer");
> >> -    if (ret)
> >> -            return;
> >> -
> >> -    ret = delete_fdt_mem_rsv(fdt, addr, size);
> >> -    if (!ret)
> >> -            pr_debug("Removed old IMA buffer reservation.\n");
> >> -}
> >> -
> >>   #ifdef CONFIG_IMA_KEXEC
> >>   /**
> >>    * arch_ima_add_kexec_buffer - do arch-specific steps to add the IMA buffer
> >>    *
> >> + * @image: kimage struct to set IMA buffer data
> >> + * @load_addr: Starting address where IMA buffer is loaded at
> >> + * @size: Number of bytes in the IMA buffer
> >> + *
> >>    * Architectures should use this function to pass on the IMA buffer
> >>    * information to the next kernel.
> >>    *
> >> @@ -179,7 +140,7 @@ int setup_ima_buffer(const struct kimage *image, void *fdt, int chosen_node)
> >>      int ret, addr_cells, size_cells, entry_size;
> >>      u8 value[16];
> >>
> >> -    remove_ima_buffer(fdt, chosen_node);
> >> +    remove_ima_kexec_buffer(fdt, chosen_node);
> >>      if (!image->arch.ima_buffer_size)
> >>              return 0;
> >>
> >> @@ -201,7 +162,7 @@ int setup_ima_buffer(const struct kimage *image, void *fdt, int chosen_node)
> >>      if (ret)
> >>              return ret;
> >>
> >> -    ret = fdt_setprop(fdt, chosen_node, "linux,ima-kexec-buffer", value,
> >> +    ret = fdt_setprop(fdt, chosen_node, FDT_PROP_IMA_KEXEC_BUFFER, value,
> >>                        entry_size);
> >>      if (ret < 0)
> >>              return -EINVAL;
> >> diff --git a/drivers/of/fdt.c b/drivers/of/fdt.c
> >> index 4602e467ca8b..d2e6f8ce0e42 100644
> >> --- a/drivers/of/fdt.c
> >> +++ b/drivers/of/fdt.c
> >> @@ -25,6 +25,7 @@
> >>   #include <linux/serial_core.h>
> >>   #include <linux/sysfs.h>
> >>   #include <linux/random.h>
> >> +#include <linux/kexec.h>
> >>
> >>   #include <asm/setup.h>  /* for COMMAND_LINE_SIZE */
> >>   #include <asm/page.h>
> >> @@ -1289,4 +1290,113 @@ static int __init of_fdt_raw_init(void)
> >>   late_initcall(of_fdt_raw_init);
> >>   #endif
> >>
> >> +#ifdef CONFIG_HAVE_IMA_KEXEC
> >
> > Can we avoid #ifdef and use IS_ENABLED() within the functions?
> I can use IS_ENABLED() or move the functions to another C file and
> conditionally build based on the CONFIG.
>
> >
> >> +/**
> >> + * do_get_kexec_buffer - Get address and size of IMA kexec buffer
> >> + *
> >> + * @prop: IMA kexec buffer node in the device tree
> >> + * @len: Size of the given device tree node property
> >> + * @addr: Return address of the node
> >> + * @size: Return size of the node
> >> + */
> >> +int do_get_kexec_buffer(const void *prop, int len, unsigned long *addr,
> >> +                    size_t *size)
> >> +{
> >> +    int addr_cells, size_cells;
> >> +    struct device_node *root;
> >> +
> >> +    root = of_find_node_by_path("/");
> >
> > The code in fdt.c operates on flat trees. This is an unflattened tree.
>
> Would it be better if I move these functions to a new C file under
> "drivers/of"?
>
> >
> >> +    if (!root)
> >> +            return -EINVAL;
> >> +
> >> +    addr_cells = of_n_addr_cells(root);
> >> +    size_cells = of_n_size_cells(root);
> >> +
> >> +    of_node_put(root);
> >> +
> >> +    if (len < 4 * (addr_cells + size_cells))
> >> +            return -ENOENT;
> >> +
> >> +    *addr = of_read_number(prop, addr_cells);
> >> +    *size = of_read_number(prop + 4 * addr_cells, size_cells);
> >
> > There's nothing in this function specific to 'kexec buffer'.
> >
> > This interface is kind of broken. 'prop' could come from anywhere in
> > the tree, but we always read the root address and size cells. Those only
> > apply to immediate child node properties. And anything other than
> > immediate root child nodes, there needs to be address translation.
> >
>
> I moved this function from arch/powerpc/kexec/ima.c to here so that it
> can be shared by ARM64 as well. Below is one of the usage of
> do_get_kexec_buffer().
>
>          prop = of_get_property(of_chosen, "linux,ima-kexec-buffer", &len);
>          if (!prop)
>                  return -ENOENT;
>
>          ret = do_get_kexec_buffer(prop, len, &tmp_addr, &tmp_size);
>
> Is there a way to detect if the 'prop' is an immediate root child node
> or not so that we can handle it appropriately in this interface function?

Just don't split up the function.

> >> +
> >> +    return 0;
> >> +}
> >> +
> >> +/**
> >> + * remove_ima_kexec_buffer - remove the IMA buffer property and
> >> + *                       reservation from @fdt
> >
> > IIRC, kerneldoc requires this to be one line.
> I'll upate.
>
> >
> >> + *
> >> + * @fdt: Flattened Device Tree to update
> >> + * @chosen_node: Offset to the chosen node in the device tree
> >> + *
> >> + * The IMA measurement buffer is of no use to a subsequent kernel,
> >> + * so we always remove it from the device tree.
> >> + */
> >> +void remove_ima_kexec_buffer(void *fdt, int chosen_node)
> >> +{
> >
> > Can't this go in some common kexec code?
>
> I'd considered moving this to "kernel", but Thiago suggested
> "drivers/of" is a better place. If you have suggestions for a better
> place, please let me know.

Humm, I guess drivers/of/ is fine, but it should be its own file.
Also, there's nothing arch specific about handling bootargs, initrd,
etc., so create a common function to handle all that. You can probably
include allocating and copying the new fdt as part of that. Powerpc
does kmalloc and arm64 does vmalloc for the new fdt. Is that a
necessary difference?

Then the IMA handling can all be private to the new file?

>
> Note that this function needs to be available even when CONFIG_IMA is
> not enabled. So I can't move it to "security/integrity/ima".
>
> >> +    int ret, len;
> >> +    unsigned long addr;
> >> +    size_t size;
> >> +    const void *prop;
> >> +
> >> +    prop = fdt_getprop(fdt, chosen_node, FDT_PROP_IMA_KEXEC_BUFFER, &len);
> >> +    if (!prop) {
> >> +            pr_debug("Unable to find the ima kexec buffer node\n");
> >> +            return;
> >> +    }
> >> +
> >> +    ret = do_get_kexec_buffer(prop, len, &addr, &size);
> >> +    fdt_delprop(fdt, chosen_node, FDT_PROP_IMA_KEXEC_BUFFER);
> >> +    if (ret) {
> >> +            pr_err("Unable to delete the ima kexec buffer node\n");
> >> +            return;
> >> +    }
> >> +
> >> +    ret = delete_fdt_mem_rsv(fdt, addr, size);
> >> +    if (!ret)
> >> +            pr_debug("Removed old IMA buffer reservation.\n");
> >> +}
> >> +#endif /* CONFIG_HAVE_IMA_KEXEC */
> >> +
> >> +#ifdef CONFIG_KEXEC_FILE
> >> +/**
> >> + * delete_fdt_mem_rsv - delete memory reservation with given address and size
> >> + *
> >> + * @fdt: Flattened Device Tree to update
> >> + * @start: Starting address of the reservation to delete
> >> + * @size: Size of the reservation to delete
> >> + *
> >> + * Return: 0 on success, or negative errno on error.
> >> + */
> >> +int delete_fdt_mem_rsv(void *fdt, unsigned long start, unsigned long size)
> >> +{
> >> +    int i, ret, num_rsvs = fdt_num_mem_rsv(fdt);
> >> +
> >> +    for (i = 0; i < num_rsvs; i++) {
> >> +            uint64_t rsv_start, rsv_size;
> >> +
> >> +            ret = fdt_get_mem_rsv(fdt, i, &rsv_start, &rsv_size);
> >> +            if (ret) {
> >> +                    pr_err("Malformed device tree.\n");
> >> +                    return -EINVAL;
> >> +            }
> >> +
> >> +            if (rsv_start == start && rsv_size == size) {
> >> +                    ret = fdt_del_mem_rsv(fdt, i);
> >> +                    if (ret) {
> >> +                            pr_err("Error deleting device tree reservation.\n");
> >> +                            return -EINVAL;
> >> +                    }
> >> +
> >> +                    pr_debug("Freed reserved memory at %lu of size %lu\n",
> >> +                             start, size);
> >> +                    return 0;
> >> +            }
> >> +    }
> >> +
> >> +    return -ENOENT;
> >> +}
> >> +#endif /* CONFIG_KEXEC_FILE */
> >> +
> >>   #endif /* CONFIG_OF_EARLY_FLATTREE */
> >> diff --git a/include/linux/kexec.h b/include/linux/kexec.h
> >> index 9e93bef52968..6c6c6791a7ba 100644
> >> --- a/include/linux/kexec.h
> >> +++ b/include/linux/kexec.h
> >> @@ -407,6 +407,30 @@ static inline int kexec_crash_loaded(void) { return 0; }
> >>   #define kexec_in_progress false
> >>   #endif /* CONFIG_KEXEC_CORE */
> >>
> >> +#if defined(CONFIG_OF_EARLY_FLATTREE) && defined(CONFIG_HAVE_IMA_KEXEC)
> >
> > CONFIG_OF_EARLY_FLATTREE is wrong because that's all early boot (i.e.
> > init section) functions.
> >
> > If these functions are implemented in fdt.c, then this is the wrong
> > header. But it's the implementation that should move.
>
> If you think defining these functions in "drivers/of" is okay, I'll move
> them to a new C file in OF.
>
> Would CONFIG_OF_FLATTREE be a better one to enable these ima kexec
> functions? "of_" and "fdt_" are the functions called by these ima kexec
> functions.

Yes.

>
> >
> >> +extern void remove_ima_kexec_buffer(void *fdt, int chosen_node);
> >> +extern int do_get_kexec_buffer(const void *prop, int len, unsigned long *addr,
> >> +                           size_t *size);
> >> +#else
> >> +static inline void remove_ima_kexec_buffer(void *fdt, int chosen_node) {}
> >> +static inline int do_get_kexec_buffer(const void *prop, int len,
> >> +                                  unsigned long *addr, size_t *size)
> >> +{
> >> +    return -EOPNOTSUPP;
> >> +}
> >> +#endif /* CONFIG_OF_EARLY_FLATTREE && CONFIG_HAVE_IMA_KEXEC */
> >> +
> >> +#if defined(CONFIG_OF_EARLY_FLATTREE) && defined(CONFIG_KEXEC_FILE)
> >> +extern int delete_fdt_mem_rsv(void *fdt, unsigned long start,
> >> +                          unsigned long size);
> >> +#else
> >> +static inline int delete_fdt_mem_rsv(void *fdt, unsigned long start,
> >> +                                 unsigned long size)
> >> +{
> >> +    return 0;
> >> +}
> >> +#endif /* CONFIG_OF_EARLY_FLATTREE && CONFIG_KEXEC_FILE */
> >> +
> >>   #endif /* !defined(__ASSEBMLY__) */
> >>
> >>   #endif /* LINUX_KEXEC_H */
> >> diff --git a/include/linux/libfdt.h b/include/linux/libfdt.h
> >> index 90ed4ebfa692..75fb40aa013b 100644
> >> --- a/include/linux/libfdt.h
> >> +++ b/include/linux/libfdt.h
> >> @@ -5,4 +5,7 @@
> >>   #include <linux/libfdt_env.h>
> >>   #include "../../scripts/dtc/libfdt/libfdt.h"
> >>
> >> +/* Common device tree properties */
> >> +#define FDT_PROP_IMA_KEXEC_BUFFER   "linux,ima-kexec-buffer"
> >
> > This is not part of libfdt. We generally don't do defines for DT
> > strings.
> Should I create a new header file for this or would you suggest using an
> existing one?

I'm suggesting no define at all. The string is already descriptive, so
what do you gain with the define name? Nothing.

Rob

Powered by blists - more mailing lists