[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAGeo-3fWSP=98a=btSVdzRBYNC1B1P91R7=L4wxNcBjbE7hMQA@mail.gmail.com>
Date: Sat, 25 Mar 2023 15:07:14 +0800
From: Hongbin Ji <jihongbin999@...il.com>
To: Mike Rapoport <rppt@...nel.org>
Cc: akpm@...ux-foundation.org, linux-mm@...ck.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH] memblock: Make memblock memblock_dbg info handle
overflowing range @base + @size
Actually @base + @size overflows
On Sat, Mar 25, 2023 at 3:04 PM Hongbin Ji <jihongbin999@...il.com> wrote:
>
> Sorry, this is the first time I use email, I checked the top post and
> bottom post just now, I will modify the sending method.
>
> Passing an oversized @size argument is allowed inside membloc_remove().
>
> static inline phys_addr_t memblock_cap_size(phys_addr_t base, phys_addr_t *size)
> {
> return *size = min(*size, (phys_addr_t)ULLONG_MAX - base);
> }
>
> phys_addr_t end = base + memblock_cap_size(base, &size);
>
> and internally checks and handles @size parameter overflow
>
> Mike Rapoport <rppt@...nel.org> 于2023年3月25日周六 14:42写道:
> >
> > On Sat, Mar 25, 2023 at 02:25:58PM +0800, Hongbin Ji wrote:
> > > It is just to correct the information displayed by the debugging.
> > > The wrong information display is also a problem, but it is not a
> > > problem that affects the function
> >
> > Please don't top post.
> >
> > Wrong debugging info will be the least of the problems if memblock_add() or
> > membloc_remove() are called with wrong parameters.
> >
> > Please work on cleanups based on code inspection outside of mm/
> >
> > > Mike Rapoport <rppt@...nel.org> 于2023年3月25日周六 14:04写道:
> > > >
> > > > On Fri, Mar 24, 2023 at 04:15:13PM +0800, 纪宏宾 wrote:
> > > > > Allow memblock users to specify range where @base + @size overflows,
> > > > > This will cause the address range information in the debug output to
> > > > > be displayed incorrectly.
> > > >
> > > > Is there a real problem you are trying to solve?
> > > >
> > > > > For example, calling memblock_remove(1ULL << PHYS_MASK_SHIFT,
> > > > > ULLONG_MAX) in arch/arm64/mm/init.c,
> > > > > would be displayed as:
> > > > > [ 0.000000] memblock_remove: [0x0001000000000000-0x0000fffffffffffe]
> > > > > arm64_memblock_init+0x24/0x270
> > > > > but we expect the output:
> > > > > [ 0.000000] memblock_remove: [0x0001000000000000-0xffffffffffffffff]
> > > > > arm64_memblock_init+0x24/0x270
> > > > >
> > > > > Signed-off-by: Hongbin Ji <jhb_ee@....com>
> > > > > ---
> > > > > mm/memblock.c | 14 +++++++-------
> > > > > 1 file changed, 7 insertions(+), 7 deletions(-)
> > > > >
> > > > > diff --git a/mm/memblock.c b/mm/memblock.c
> > > > > index 25fd0626a9e7..567b99e4355d 100644
> > > > > --- a/mm/memblock.c
> > > > > +++ b/mm/memblock.c
> > > > > @@ -700,7 +700,7 @@ static int __init_memblock
> > > > > memblock_add_range(struct memblock_type *type,
> > > > > int __init_memblock memblock_add_node(phys_addr_t base, phys_addr_t size,
> > > > > int nid, enum memblock_flags flags)
> > > > > {
> > > > > - phys_addr_t end = base + size - 1;
> > > > > + phys_addr_t end = base + min(size, PHYS_ADDR_MAX - base + 1) - 1;
> > > > >
> > > > > memblock_dbg("%s: [%pa-%pa] nid=%d flags=%x %pS\n", __func__,
> > > > > &base, &end, nid, flags, (void *)_RET_IP_);
> > > > > @@ -721,7 +721,7 @@ int __init_memblock memblock_add_node(phys_addr_t
> > > > > base, phys_addr_t size,
> > > > > */
> > > > > int __init_memblock memblock_add(phys_addr_t base, phys_addr_t size)
> > > > > {
> > > > > - phys_addr_t end = base + size - 1;
> > > > > + phys_addr_t end = base + min(size, PHYS_ADDR_MAX - base + 1) - 1;
> > > > >
> > > > > memblock_dbg("%s: [%pa-%pa] %pS\n", __func__,
> > > > > &base, &end, (void *)_RET_IP_);
> > > > > @@ -822,7 +822,7 @@ static int __init_memblock
> > > > > memblock_remove_range(struct memblock_type *type,
> > > > >
> > > > > int __init_memblock memblock_remove(phys_addr_t base, phys_addr_t size)
> > > > > {
> > > > > - phys_addr_t end = base + size - 1;
> > > > > + phys_addr_t end = base + min(size, PHYS_ADDR_MAX - base + 1) - 1;
> > > > >
> > > > > memblock_dbg("%s: [%pa-%pa] %pS\n", __func__,
> > > > > &base, &end, (void *)_RET_IP_);
> > > > > @@ -854,7 +854,7 @@ void __init_memblock memblock_free(void *ptr, size_t size)
> > > > > */
> > > > > int __init_memblock memblock_phys_free(phys_addr_t base, phys_addr_t size)
> > > > > {
> > > > > - phys_addr_t end = base + size - 1;
> > > > > + phys_addr_t end = base + min(size, PHYS_ADDR_MAX - base + 1) - 1;
> > > > >
> > > > > memblock_dbg("%s: [%pa-%pa] %pS\n", __func__,
> > > > > &base, &end, (void *)_RET_IP_);
> > > > > @@ -865,7 +865,7 @@ int __init_memblock memblock_phys_free(phys_addr_t
> > > > > base, phys_addr_t size)
> > > > >
> > > > > int __init_memblock memblock_reserve(phys_addr_t base, phys_addr_t size)
> > > > > {
> > > > > - phys_addr_t end = base + size - 1;
> > > > > + phys_addr_t end = base + min(size, PHYS_ADDR_MAX - base + 1) - 1;
> > > > >
> > > > > memblock_dbg("%s: [%pa-%pa] %pS\n", __func__,
> > > > > &base, &end, (void *)_RET_IP_);
> > > > > @@ -876,7 +876,7 @@ int __init_memblock memblock_reserve(phys_addr_t
> > > > > base, phys_addr_t size)
> > > > > #ifdef CONFIG_HAVE_MEMBLOCK_PHYS_MAP
> > > > > int __init_memblock memblock_physmem_add(phys_addr_t base, phys_addr_t size)
> > > > > {
> > > > > - phys_addr_t end = base + size - 1;
> > > > > + phys_addr_t end = base + min(size, PHYS_ADDR_MAX - base + 1) - 1;
> > > > >
> > > > > memblock_dbg("%s: [%pa-%pa] %pS\n", __func__,
> > > > > &base, &end, (void *)_RET_IP_);
> > > > > @@ -1645,7 +1645,7 @@ void __init memblock_free_late(phys_addr_t base,
> > > > > phys_addr_t size)
> > > > > {
> > > > > phys_addr_t cursor, end;
> > > > >
> > > > > - end = base + size - 1;
> > > > > + end = base + min(size, PHYS_ADDR_MAX - base + 1) - 1;
> > > > > memblock_dbg("%s: [%pa-%pa] %pS\n",
> > > > > __func__, &base, &end, (void *)_RET_IP_);
> > > > > kmemleak_free_part_phys(base, size);
> > > > > --
> > > > > 2.34.1
> > > >
> > > > --
> > > > Sincerely yours,
> > > > Mike.
> >
> > --
> > Sincerely yours,
> > Mike.
Powered by blists - more mailing lists